You are on page 1of 304

Report Number: C4-040R-02

Router Security Configuration Guide
Principles and guidance for secure configuration of IP routers, with detailed instructions for Cisco Systems routers

Router Security Guidance Activity of the System and Network Attack Center (SNAC)
Authors: Vanessa Antoine Raymond Bongiorni Anthony Borza Patricia Bosmajian Daniel Duesterhaus Michael Dransfield Brian Eppinger Kevin Gallicchio Stephen Hamilton James Houser Andrew Kim Phyllis Lee Brian McNamara Tom Miller David Opitz Florence Richburg Michael Wiacek Mark Wilson Neal Ziring December 15, 2005 Version: 1.1c

National Security Agency 9800 Savage Rd. Suite 6704 Ft. Meade, MD 20755-6704

Router Security Configuration Guide

This document is only a guide to recommended security settings for Internet Protocol (IP) routers, particularly routers running Cisco Systems Internet Operating System (IOS) versions 11.3 through 12.4. It cannot replace well-designed policy or sound judgment. This guide does not address site-specific configuration issues. Care must be taken when implementing the security steps specified in this guide. Ensure that all security steps and procedures chosen from this guide are thoroughly tested and reviewed prior to imposing them on an operational network.

This document is current as of October, 2005. The most recent version of this document may always be obtained through

The authors would like to acknowledge Daniel Duesterhaus, author of the original NSA “Cisco Router Security Configuration Guide,” and the management and staff of the Applications and Architectures division for their patience and assistance with the initial development of this guide. Special thanks also go to Ray Bongiorni for quality assurance and editorial work, and to Julie Martz and Kathy Jones for proof-reading assistance. Additional contributors to the guide effort include Andrew Dorsett, Charles Hall, Scott McKay, and Jeffrey Thomas. Thanks must also be given to the dozens of professionals outside NSA who made suggestions for the improvement of this document, especially George Jones, John Stewart, and Joshua Wright.

Trademark Information
Cisco, IOS, and CiscoSecure are registered trademarks of Cisco Systems, Inc. in the USA and other countries. Windows 2000 and Windows XP are registered trademarks of Microsoft Corporation in the USA and other countries. All other names are trademarks or registered trademarks of their respective companies.

Revision History
1.0 1.0b 1.0f 1.0g 1.0h 1.0j 1.0k 1.1 1.1b 1.1c Sep 2000 Oct 2000 Mar 2001 Apr 2001 Aug 2001 Nov 2001 Mar 2002 Sep 2002 Dec 2003 Dec 2005 First complete draft, extensive internal review. Revised after review by Ray Bongiorni Second release version: second pre-pub review Third release version: incorporated external feedback. Fourth release version; another QA review. Fifth release version. Last release of 1.0, another pre-pub review. Major revision and expansion, another pre-pub review Minor revision, corrections, additions, fixed links Updated, fixed inconsistencies, checked links


Version 1.1c


Preface 1. Introduction
1.1. 1.2. 1.3. 1.4.

5 7

The Roles of Routers in Modern Networks .....................................................................7 Motivations for Providing Router Security Guidance......................................................9 Typographic and Diagrammatic Conventions Used in this Guide .................................10 Structural Overview .......................................................................................................12

2. Background and Review
2.1. 2.2. 2.3. 2.4. 2.5. 2.6. 2.7.


Review of TCP/IP Networking ......................................................................................15 TCP/IP and the OSI Model ............................................................................................17 Review of IP Routing and IP Architectures ...................................................................19 Basic Router Functional Architecture ............................................................................24 Review of Router-Relevant Protocols and Layers .........................................................27 Quick “Review” of Attacks on Routers .........................................................................29 References......................................................................................................................30

3. Router Security Principles and Goals
3.1. 3.2. 3.3. 3.4. 3.5.


Protecting the Router Itself ............................................................................................33 Protecting the Network with the Router.........................................................................35 Managing the Router......................................................................................................43 Security Policy for Routers ............................................................................................46 References......................................................................................................................51

4. Implementing Security on Cisco Routers
4.1. 4.2. 4.3. 4.4. 4.5. 4.6. 4.7.


Router Access Security ..................................................................................................55 Router Network Service Security...................................................................................70 Access Control Lists, Filtering, and Rate Limiting........................................................83 Routing and Routing Protocols ....................................................................................102 Audit and Management................................................................................................139 Security for Router Network Access Services .............................................................175 Collected References....................................................................................................202

5. Advanced Security Services
5.1. 5.2. 5.3. 5.4. 5.5. 5.6.


Role of the Router in Inter-Network Security ..............................................................204 IP Network Security.....................................................................................................205 Using SSH for Remote Administration Security .........................................................227 Using a Cisco Router as a Firewall ..............................................................................232 Cisco IOS Intrusion Detection .....................................................................................241 References....................................................................................................................247

6. Testing and Security Validation


Version 1.1c


Router Security Configuration Guide

6.1. 6.2. 6.3. 6.4. 6.5.

Principles for Router Security Testing .........................................................................250 Testing Tools................................................................................................................250 Testing and Security Analysis Techniques ..................................................................251 Using the Router Audit Tool........................................................................................258 References....................................................................................................................261

7. Additional Issues in Router Security
7.1. 7.2. 7.3. 7.4. 7.5. 7.6. 7.7. 7.8. 7.9.


Routing and Switching.................................................................................................263 IPv6 ..............................................................................................................................265 ATM and IP Routing....................................................................................................266 Multi-Protocol Label Switching (MPLS).....................................................................267 IPSec and Dynamic Virtual Private Networks .............................................................268 Tunneling Protocols and Virtual Network Applications ..............................................269 IP Quality of Service (QoS) and RSVP........................................................................270 Secure DNS..................................................................................................................271 References....................................................................................................................272

8. Appendices
8.1. 8.2. 8.3. 8.4.


Top Ways to Quickly Improve the Security of a Cisco Router....................................274 Application to Ethernet Switches and Related Non-Router Network Hardware..........280 Overview of Cisco IOS Versions and Releases ...........................................................283 Glossary of Router Security-related Terms..................................................................289

9. Additional Resources
9.1. 9.2. 9.3.


Bibliography.................................................................................................................295 Web Site References ....................................................................................................298 Tool References ...........................................................................................................300




Version 1.1c


Routers direct and control much of the data flowing across computer networks. This guide provides technical guidance intended to help network administrators and security officers improve the security of their networks. Using the information presented here, you can configure your routers to control access, resist attacks, shield other network components, and protect the integrity and confidentiality of network traffic. This guide was developed in response to numerous questions and requests for assistance received by the NSA System and Network Attack Center (SNAC). The topics covered in the guide were selected on the basis of customer interest, community concensus, and the SNAC’s background in securing networks. The goal for this guide is a simple one: improve the security provided by routers in US Government operational networks.

Who Should Use This Guide
Network administrators and network security officers are the primary audience for this configuration guide, throughout the text the familiar pronoun “you” is used for guidance directed specifically to them. Most network administrators are responsible for managing the connections within their networks, and between their network and various other networks. Network security officers are usually responsible for selecting and deploying the assurance measures applied to their networks. For this audience, this guide provides security goals and guidance, along with specific examples of configuring Cisco routers to meet those goals. Firewall administrators are another intended audience for this guide. Often, firewalls are employed in conjunction with filtering routers; the overall perimeter security of an enclave benefits when the configurations of the firewall and router are complementary. While this guide does not discuss general firewall topics in any depth, it does provide information that firewall administrators need to configure their routers to actively support their perimeter security policies. Section 5 includes information on using the firewall features of the Cisco Integrated Security facility. Information System Security Engineers (ISSEs) may also find this guide useful. Using it, an ISSE can gain greater familiarity with security services that routers can provide, and use that knowledge to incorporate routers more effectively into the secure network configurations that they design. Sections 4, 5, and 6 of this guide are designed for use with routers made by Cisco Systems, and running Cisco’s IOS software. The descriptions and examples in those sections were written with the assumption that the reader is familiar with basic Cisco router operations and command syntax.

Version 1.1c


Router Security Configuration Guide

This guide was created by a team of individuals in the System and Network Attack Center (SNAC), which is part of the NSA Information Assurance Directorate. The editor was Neal Ziring. Comments and feedback about this guide may be directed to the SNAC (Attn: Neal Ziring), Suite 6704, National Security Agency, Ft. Meade, MD, 20755-6704, or via e-mail to


Version 1.1c

Directing data between portions of a network is the primary purpose of a router. Router 1 forwards the packet to Router 2.2. below. Consulting its internal route table. See Section 2. and sends the packet over LAN 1 to its gateway.10 Figure 1-1 – A Simple Network with Two Routers If the user host (top left) needs to send a message to the file server (bottom right). The Roles of Routers in Modern Networks On a very small computer network. Router 1. In larger.2.10. Therefore.0 File Server 14. Introduction 1. most modern networks cannot function. Figure 1-1.2.20. data must be directed specifically to the intended destination.2. Consulting its own route table. it is feasible to use simple broadcast or sequential mechanisms for moving data from point to point.Introduction 1.12 Router 1 Wide Area Network LAN 2 14. the operation of any large network depends on the route tables in all of its constituent routers.9. In practice. Version 1. or packets. based on internal addresses and tables of routes.1c 7 . An Ethernet local area network (LAN) is essentially a broadcast network. illustrates the primary function of a router in a small IP network.3 for a quick review of TCP/IP and IP addressing. or known destinations that serve certain addresses. the security of routers and their configuration settings is vital to network operation. it creates a packet with address 14. Most large computer networks use the TCP/IP protocol suite.0 User Host 190. Router 2 sends the packet over LAN 3 to the File Server.20.0 Router 2 LAN 3 14. Routers direct network data messages.2.9.2. more complex networks. Without robust routing. LAN 1

1c .Router Security Configuration Guide In addition to directing packets. 8 Version 1. a router may be responsible for filtering traffic. it allows them to protect computers and other network components from illegitimate or hostile traffic. allowing some data packets to pass and rejecting others. and 5. Filtering is a very important responsibility for routers. 4. consult Sections 3. For more information.

denial of network communication services. • On the other hand. ensure data integrity. with pertinent references at the end of each section identifying the most useful vendor documentation. and do not always supply security rationale for selecting and applying those features. Cisco offers extensive on-line documentation and printed books about the security features supported by their products. • Compromise of a router’s access control can result in exposure of network configuration details or denial of service. and available software.2. and make it easier for attackers to avoid detection. These books and papers are valuable. thus avoiding a very serious potential source of security problems. • A poor router filtering configuration can reduce the overall security of an entire enclave. and exposure of sensitive data. and facilitate secure cooperation between independent enclaves. vendor documents. This guide also provides pointers to related books. Version 1.Introduction 1. This guide attempts to provide security rationale and concrete security direction. In general.1c 9 . standards. There are substantial security resources available from router vendors. Security policy enforced at a router is difficult for negligent or malicious end-users to circumvent. proper use of router cryptographic security features can help protect sensitive data. expose internal network components to scans and attacks. Compromise of a router can lead to various security problems on the network served by that router. but they are not sufficient. • Compromise of a router’s route tables can result in reduced performance. Most vendor-supplied router security documents are focused on documenting all of the security features offered by the router. well-configured secure routers can greatly improve the overall security posture of a network. or even other networks with which that router communicates. Motivations for Providing Router Security Guidance Routers provide services that are essential to the correct. secure operation of the networks they serve. and can facilitate attacks against other network components. For example.

! set the log host IP address and buffer size logging 14. usually a remark about the line that follows it.B. to make it narrower.) • Cisco IOS accepts the shortest unique. Input in the transcript is distinguished from output. the interface name ethernet is commonly abbreviated “eth” and the command configure terminal is commonly abbreviated “config t”. using Courier typeface.9.Router Security Configuration Guide 1. The following typographic conventions are used as part of presenting the examples.” • Sequences of commands to be used in a configuration are shown separately from the text.B. .6 logging buffered 16000 • Transcripts of router sessions are shown separately from the text. output that would be too wide to fit on the page is shown with some white space removed. Central# exit • IP addresses will be shown in the text and in diagrams as A.” Command arguments are shown in Courier italics: “syntax for a simple IP access list rule is access-list number permit host address. this classless netmask notation will be used where a netmask is relevant.9. the bare address will be used.255.2. use the command show ip route.250 YES Manual . In some cases.D/N. this guide uses many abbreviations commonly employed in the Cisco documentation and literature. 10 Version 1.15.C.250 YES NVRAM Ethernet 0/1 where N is the number of set bits in the IP netmask.2. For example. Central> enable Password: Central# ! list interfaces in concise format Central# show ip interface brief Interface IP Address OK? Method Ethernet 0/0 14.150/24 has a netmask of 255.C. user input and comments are shown in Courier bold typeface. • Specific router and host commands are identified in the text using Courier bold typeface: “to list the current routing table. Typographic and Diagrammatic Conventions Used in this Guide To help make this guide more practical. most of the sections include extensive instructions and examples. (In general. Otherwise.3. For commands that are typed very frequently.9.1c . using Courier typeface. The exclamation point begins a comment line. For example. 14. unambiguous abbreviation for any command or keyword. or as A. Elision of long output is denoted by two dots.

0/24 A local-area network (LAN) segment. This guide uses the following set of icons in all of its diagrams. This icon represents a router.1c 11 . Each router is presumed to have an administrative console line connection. and may include bridges. they are shown broken across several lines.34. Computers on the network are represented with one of these two icons. and other network hardware. link encrypters. such as an Ethernet. The IOS command line interface will not permit this. with several connections. is represented by a horizontal or vertical bus. Each line connected to a router icon represents a network interface on that router.Introduction • In a few cases. Workstation Server Router2 Small LAN 12. Such networks normally include other routers.56. Discussions of network structure and security frequently depend on network diagrams. switches. This icon represents a LAN or a wide-area network over which routers communicate. which is not shown. commands shown in examples are too long to fit on one line. you will need to type the long command on one line. when attempting to apply these examples. Network Version 1.

• Section 4 details the methods and commands for applying security to Cisco routers. along with a discussion of relationships between router security and overall network security. • Section 9 provides a list of resources. • Section 8 consists of four diverse appendices: • • • • tips for quickly improving the security of a router how to apply parts of this guide to LAN switches overview of the Cisco IOS software family and versions. The main topics of this section are IP security (IPSec). including pointers to web sites and security tools. The list below describes the major sections. collected from all the sections of the guide. controlling traffic and filtering using a router. • Section 6 presents testing and troubleshooting techniques for router security. • Section 3 presents a security model for routers. security management for routers.4. with a focus on Cisco routers’ capabilities. • Section 2 reviews some background information about TCP/IP networking and network security. but which may become important in the near future.1c . using recent versions of the Cisco IOS software. this section describes some areas that a router security policy should address. 12 Version 1. • Section 5 describes advanced security services that some routers can provide. Structural Overview The various parts of this guide are designed to be fairly independent. configuring routing protocols security. and a router security glossary.Router Security Configuration Guide 1. Secure Shell (SSH). Security mechanisms must be applied in support of security policy. and this section presents both vendorindependent and Cisco-specific testing techniques. It is essential for good security that any router security configuration undergoes testing. and network access control for routers. It is divided into six main parts: • • • • • • securing access to the router itself. securing router network services. and describes some simple network security threats. and defines general goals and mechanisms for securing routers. readers may want to skip directly to the sections most immediately useful to them. References are included at the end of each section. • Section 7 previews some security topics that are not yet crucial for router configuration. and using a Cisco router as a simple firewall and Intrusion Detection System (IDS).

1 – for quick tips that will greatly improve router security Section 4. network. and each may need some information about router security.1c 13 .1 – for advice for quickly securing a Cisco router Section 8.Introduction How to Use This Guide Several different roles are involved in securing a network. the high-level view of router security is more important than the details of Cisco router commands. Read the sections listed below if your role is security planner or security designer. • • • • Section 8.4 – for routing protocol security instructions (unless the routers are using static routes exclusively) Version 1. For network security planners and system security designers.2 – for instructions on applying this guide to LAN switches Section 8. and other advanced security capabilities Section 7 – for a preview of potential future issues For network administrators involved in the daily operation of a network with Cisco routers. and router operational concepts Section 3 – for general router security principles Section 4. this guide offers detailed advice and direction. 5. The paragraphs below offer roadmaps for using this guide for several different network security roles. the detailed instructions for locking down a router are the most important part of this guide.3 – for a quick guide to using SSH for Cisco administration Section 8. • • • • • Section 2 – for a review of TCP/IP. Read the sections listed below if your goal is to quickly lock down a router.3 – for an idea of what Cisco routers can do for network security Section 5 – for information about Cisco router VPN.3 – for information on Cisco IOS versions and upgrades Section 9 – for an overview of recommended references and tools For network security analysts or administrators trying to improve the security posture of a network as quickly as possible. Read the sections listed below if your role is network administrator. • • • • • • • • • Section 2 – for a review.2 – for instructions on configuring IPSec on Cisco routers Section 5.1 through 4. if necessary Section 3 – for the security principles behind the advice in Section 4 Section 4 – for detailed instructions on configuring Cisco routers Section 5.1. firewall.1 – for explicit directions on router access security Section 4.3 – for advice and guidance on setting up filtering Section 4.

be sure to test it thoroughly in a lab or testbed network.Router Security Configuration Guide Before applying any of the guidance in this guide to operational routers.1c . and applying configuration changes to a router can instantly affect large numbers of hosts. In most cases. This guide provides security guidance for a large number of topics. 14 Version 1. Every section includes references to books. web sites. it is not practical for this document to include full background and technical details. and standards that you can use to obtain more information or greater detail. Operational networks are complex.

1) standard. Within a LAN. Standards that relate to LANs are primarily the IEEE 802. If you are still lost at the end of Section 2.” consider that a 10BaseT Ethernet (10 megabit per second using twisted pair cabling) has a usual maximum of 1024 stations over a maximum cable distance of 2500 meters.1. data must be directed more carefully. either by preventing them or detecting them and taking appropriate action. In almost all cases.1c 15 . If you are lost at this point. keep reading because much of this will be explained below. 2. then consider reading parts of some of the books and/or web pages listed at the end of the section. In order to keep this section brief. and network security. Just above the MAC level is the Logical Link Control (802. the reference list at the end of the section includes a long list of other useful sources of background information. it is feasible to use simple broadcast or sequential (token) mechanisms for moving data from point to point.2) standard and above that it the High Level Interface (802. it glosses over a lot of issues. Purpose of a Router In larger. As one would expect. Token based LANs avoid collisions by only allowing one host at time to transmit (the host that currently has the token may transmit).5 is the MAC standard for Token Ring. broadcast LANs must deal with collisions. A local area network is composed of a relatively small number of hosts connected over a relatively small physical area. For instance. This is where routers come in. 2. router hardware architecture. To compensate for that briefness. Background and Review This section reviews some background information about TCP/IP networking.1. more complex computer networks. An Ethernet local area network (LAN) is essentially a (logical) bus based broadcast network. Between LANs using TCP/IP.Background and Review 2. router software architecture. Readers with a good grasp of network and router fundamentals may want to skip this section.1. Review of TCP/IP Networking As mentioned in Section 1. 802. large networks are actually composed of a collection of LANs that are interconnected or “internetworked”. For instance. Routers take Version 1. on a small computer network. To give some meaning to the term “relatively. addressing is done with a MAC address. while 802. “Relatively small” is the important phrase here. a typical office LAN. might have 100 computers (and printers) attached to a switch or set of hubs.1.3 is the Media Access Control (MAC) standard for CSMA/CD (the Ethernet standard). though the physical implementation may use hubs (with a physical star topology). but since it is relatively brief. why not humor the author and read on. addressing is done using IP addresses.x series. using 100BaseT Ethernet.

the router learns about where various addresses are relative to itself and builds up route tables based on this information. respectively. not at the contents or data in the packet. based on: the addresses contained within the packet headers and a table of routes maintained within the router.2. Most large computer networks use the TCP/IP protocol suite. See Section 2. For more information about filtering. In dynamic routing. Other routers will generally only look at a packet’s header information. that a message passes through.1. see Section 3 and Section 4. There are a number of schemes or routing protocols for routers to acquire and share route table information. In the past these tables were built and updated by hand and this is referred to as static routing. one of tasks of a router is to maintain route tables which are used to decide where a packet is to go and thus which interface it should be sent out. In addition to directing packets.4.Router Security Configuration Guide network data messages from a LAN and convert them into packets suitable for transmission beyond the LAN on a wide area network (WAN). 16 Version 1. While a thorough treatment of the details is beyond the scope of this document. The goal is almost always to get these packets to another LAN and ultimately to the correct host on that LAN. Part of the “conversion” process is to add a packet header. it allows them to help protect computers and other network components. a router may be responsible for filtering traffic. Updating these routing tables and forwarding data packets between portions of a network are two of the primary tasks of a router.2 for a quick review of TCP/IP and IP addressing. Building packets and unwrapping packets are additional router functions performed by the first and last routers. There is no reason that routers cannot be used to send messages between hosts (as shown in Figure 1-1) but more typically routers are used to connect LANs to each other or to connect a LAN to a WAN. It is also possible that at the destination end a router may have to break large packets up to accommodate the size limits of the destination LAN. allowing some packets to pass through and rejecting others. Route Tables As mentioned. Routers also make decisions about where to send these packets.1c . 2. there is a substantial discussion of routing protocols is in Section 4. Filtering can be a very important function of routers. In some sense this is the lingua franca of the Internet.

2.2.1. The OSI Model After TCP/IP was well-established and other networking protocols. or “Request for Comments” issued by the Internet Engineering Task Force (IETF) as described in Section 2. Note that the word “Defense” has been deleted and added back over time. Also. Network Layers and Standards. The definitive source for information on TCP/IP are the RFCs. See Figure 2-1. for a visual mapping of TCP/IP to the Version 1. and compression of data.2. The Defense Advanced Research Projects Agency (DARPA) originated TCP/IP. some routers can be used to “translate” between different protocols running on either side of themselves.Background and Review 2. Since the development of TCP/IP preceded the ISO OSI seven layer model. Layer 4: Transport Layer deals with end to end error recovery and delivery of complete messages. TCP/IP and the OSI Model 2. so in the interest of space they are merely enumerated here. Layer 3: Network Layer deals with transmission of packets and establishing connections. Origin of TCP/IP The Transmission Control Protocol (TCP) and Internet Protocol (IP) comprise what is often seen written as TCP/IP. the Internet essentially runs on TCP/IP protocols.7. As stated above. Layer 2: Data Link Layer deals with transmission of packets on one given physical link. the International Standardization Organization (ISO) developed the Open Systems Interconnection (OSI) seven layer reference model. These seven layers are described in almost every reference. ARPA and DARPA are one and the same organization. Layer 1: Physical Layer deals with transmission of a bit stream and definition of physical link. encryption. Layer 6: Presentation Layer deals with formatting.3. Layer 5: Session Layer deals with setup and management of sessions between applications. Note that in addition to TCP/IP there are other protocols such as Novell’s IPX (Internetwork Packet eXchange) that can be used with routers.1c 17 . Layer 7: Application Layer deals with services such as email and file transfer. The National Science Foundation (NSF) also contributed to the foundation of the Internet by taking the DARPA technology and making it available to universities. such as DECnet and Novell’s IPX were operational. 2.2. the “mapping” of TCP and IP to the seven layer model is only an approximation.

the Network Layer.3 (Ethernet) Application Transport Network TCP or UDP IP Data link { Logical Link Control Media Access Control Physical Figure 2-1: Network Layers and Standards Routing occurs at layer three. A collection of various compatible protocol layers is referred to as a stack. 18 Version 1. and some of this is discussed in the following sections.1 802.Router Security Configuration Guide OSI model. This involves such details as signal voltage and duration.2 802. Layer 7 6 5 4 3 2 1 ISO/OSI Model IEEE Standards TCP/IP Model Application Presentation Session Transport Network Data Link Physical 802.1c . It also covers the mechanical aspects of connectors and cables. the Physical Layer is at a level of detail well below the concerns of this document. To fully understand routing it is useful to appreciate some of what goes on beneath it at the Data Link Layer. It is concerned with the transmission of an unstructured bit stream over a physical link. However. or optical signaling details for fiber. It may also cover some low level error control.

How far up the stack it goes depends on whether that device is the ultimate recipient of the data or merely an intermediate device. Each device on the LAN listens for packets with its address.2 Logical Link Control standard. TCP/IP. 802. For instance.4 is token bus. Within a LAN everything will be handled by Media Access Control (MAC) addresses and by a LAN protocol such as Ethernet.. is the IEEE 802. or IP addresses. in reality bits can only move from one device to another at the Physical Layer.1c 19 . but still within the OSI Data Link Layer. Figure 2-2 illustrates this process. most protocols are defined by Institute of Electrical and Electronics Engineers (IEEE) standards. routers. there is generally no need for routing. At this point it is transferred to another device and works its way up the protocol stack at that point. and which was configured with a MAC address. The MAC address is appended to the front of the data before it is placed on the LAN. Note that the data may pass through many intermediate devices on its way from the sending host to the ultimate recipient. Sending Host Intermediate Network Infrastructure Devices Receiving Host Router 1 Router n . Figure 2-2: Moving Data through Protocol Stacks Version 1. For example.3 is the Ethernet (CSMA/CD) standard. IEEE 802. Once a message is destined to leave one LAN bound for a trip across a wide area network (WAN) to another LAN.5 is token ring. Above the MAC standards. Each device on the LAN will have a something like a network interface card (NIC) which has a unique MAC address. and 802. on an Ethernet LAN each device has an appropriate Ethernet card which complies with a particular link layer standard. At this level. While one can envision logical connections at various layers in a protocol stack. do not worry about it. The IEEE 802.1 High Level Interface standard corresponds to part of the OSI Network Layer. such as 100BaseTx. it must use an IP address. it’s not essential to an understanding of routers.Background and Review 2. data begins at an application relatively high up in a protocol stack and works its way down the stack to the physical layer.3.. Thus. What is important to keep in mind is that MAC addresses are used within a LAN. Review of IP Routing and IP Architectures If one is dealing only with a local area network (LAN). If this seems confusing.

(On some newer devices it is possible to change them but normally this should not be done.1c . MAC addresses are used within a LAN by layer two (data link) protocols. 24 bits uniquely identify the manufacturer and 24 bits act as a serial number to uniquely identify the unit.Router Security Configuration Guide On the way down the stack.1. Both systems are explained below. each layer adds a relevant header to the packet. Figure 2-3 should help you visualize how headers are added. under a scheme called “Classless Inter-Domain Routing” (CIDR). all network interface devices have globally unique addresses unless their PROMs have been rewritten. They are used by layer three devices such as routers. a new header that contains a MAC address will be added in front of the IP header. 2. also sometimes called the Media Access layer. Up until the mid-1990s. Application Layer View Application Data Application Byte Stream Transport Layer View TCP Header bytes TCP (or UDP) Packet Network Layer View IP Header bytes IP Packet Media Access Layer View Media Header bytes Media Trailer optional Ethernet Packet (or other media format message) Figure 2-3: Wrapping Lower Level Headers around Data 2.2. At the data link layer. 20 Version 1. Unlike MAC addresses. Each address is unique and fixed to a particular piece of hardware. the IP header added will contain the destination IP address (in addition to other information). At the network layer. Each new header is added in front of all higher layer headers. a header will be removed at each layer.3. Ethernet MAC addresses) are 48 bits long. They are assigned by the device (or interface card) manufacturer. The header is named for the protocol layer that adds it. Still. IP addresses used a simple fixed hierarchy based on classes. MAC Addresses MAC addresses (specifically.3. Some manufacturers have had more than one identification number (more than one block of serial numbers). IP addresses are 32 bits long.) As stated previously. Also. Traditionally. IP addresses are hierarchical. today all IP address allocation on the Internet is done using masks and aggregation. On the way up the stack. IP Addresses Under the current IP version 4 standard. due to mergers and acquisitions the manufacturer identification is not as “clean” as it once was.

For a detailed explanation of TCP/IP packet header formats. referred to as Classes A.1. and D. CIDR avoids this problem by eliminating the notion of a ‘class’.3. The netmask simply specifies the number of bits in the assigned address that designate the network portion. most notably the exhaustion of Class B networks.) Version 1. Special addresses are used for such things as to broadcast to all hosts on a network or to specify a loopback packet which will never leave the host. there is a good bit of information in an IP header. B. Moreover the second 4 bits specify the length of the header. In addition to both source and destination addresses. Allocating an entire Class B network to an organization that only needed 1000 addresses would be (and was) terribly wasteful. In addition there a number of special addresses.3. an organization that needed 1000 addresses would be assigned a netmask of 22 bits. Thus it is quite feasible to introduce longer IP addresses. The class is determined by the first one to four bits of the address. the remaining bits are the host (or subnet) portion. it is also possible to take the host portion of the address and further divide that range into two fields. under CIDR. very few have enough to “fill” a Class B address range (65. there are four “classes” of IP addresses. Classful IP Addressing Under the original IP standards. which specify a subnet address and a host address respectively. This is done by specifying a parameter called a subnet mask.4. Classless Inter-Domain Routing (CIDR) and IP Addressing As the Internet grew over the 1980s and early 1990s. The number of contiguous Class C addresses allocated is a function of the size of the network. For any class. C. One was address depletion. and the other was increased route table sizes. Any address beginning with bits 110 is Class C. For example. these addresses should not be used on the Internet or other wide-area networks (see Section 4. Any address beginning with a zero bit is a Class A address.1c 21 . (Another way to think of this is that CIDR allocates several contiguous Class C addresses to a network. Any address beginning with bits 10 is a Class B address. While many networks have more hosts than a single Class C address can accommodate (255 hosts).3). The class determines how much of the 32 bit address is used to specify the network address and how much is used to specify the host within that network. It should be noted that the first 4 bits of an IP header contain a version number so new versions of the protocol can be implemented. 2. There are also a set of IP addresses that are reserved for experimental or private networks.535 hosts).7. it encountered two problems related to the expanding number of networks and hosts. and allocating a block of addresses using a netmask of the smallest size that satisfies the needs of the recipient. and any beginning with bits 1110 is class D. For a fuller discussion of subnetting see Albritton’s book [1] or one of the other references listed in Section 2. see Stevens’ book [10].Background and Review 2.3.

3. The network aggregation required by CIDR required the development of a revised routing protocol for the Internet: BGP version 4 (BGP-4).0.255) 16777216 addresses 524288 addresses Major ISP 14. They might allocate part of that large space as a block with netmask of 13 bits (524288 addresses) to a large ISP.0/18 (14.0/8 (14.14.14. The addresses of customers X and Y would still be within the large block ‘owned’ by the major allocation authority.0 .2.0/13 (14.2.0. or consult RFCs 1771 and 1772. For these topics.4 of this guide covers security issues for routing protocols. how routers exchange this information.0.0 .255.7. adding a bit more complexity to routing.14.255) 16384 addresses 16 addresses Customer Y 14. and how they decide where to forward any given packet.1c .24.1. The prefixes which specify networks under CIDR vary in length.Router Security Configuration Guide CIDR also permits address allocation authorities to allocate blocks of addresses smaller than a Class C network.63.16/28 (14.3. any meaningful discussion of routing protocols in general and BGP routing in particular is beyond the scope of this Background and Review section.255. In general. This essentially forms “supernets”.0. then they might be assigned a netmask of 28 bits.255) Customer X 14.2.3. For example.0. Aggregation To avoid explosive growth in the size of routing tables as the Internet grows it is desirable to somehow group or aggregate related network addresses together so that they form only one routing table entry. Regional Authority 14. which are composed of several related networks which are collectively advertised as only one aggregated path to that supernet. The ISP might give big customer X a block with netmask of 18 bits.0 .16-14.24. there is a detailed treatment in Huitema’s book [12]. and smaller customer Y a block with netmask of 28 bits. A major allocation authority might obtain a block of addresses with a netmask of 8 bits (16777216 addresses).255. if an organization required only 10 addresses. This reduces the number of entries required in the 22 Version 1. This is illustrated below.0.0. Another important aspect of CIDR is that it is hierarchical.24.0. the definitions of standard routing protocols specify many of the details of: how routers keep track of available routes (route tables).31) Figure 2-4: Hierarchical IP Address Range Assignment under CIDR Internet Routing and Aggregation As alluded to in Section 2.0.2. Section 4.0.

In the meantime.Background and Review route tables of routers which are “far” from a given network. so that near and far do not necessarily relate to physical distances. Beyond CIDR Until IPv6. The aggregation strategy may be based on regions (geography) or providers (network topology). is put into common use these problems will continue. with its longer addresses. CIDR has enabled the Internet community to sidestep the Class B exhaustion problem. As traffic gets to the routers “near” a given supernet more detailed routing information becomes available. CIDR and BGP-4 have helped to mitigate the problem of route table size explosion. Version 1.1c 23 .

and more thoroughly tested than a general purpose OS. But there are also a number of security benefits to a special purpose router.1c . and can also be used to partition networks and prohibit or restrict access to certain networks or services. Basic Router Functional Architecture 2. ROM is used to store a router’s bootstrap software. These are listed roughly in order of volatility. floppy drive. Flash 24 Version 1. various kinds of memory. ease of reprogramming. NVRAM.4. CPU speed and memory size are important considerations for both performance and capabilities (e.2. a specialized router operating system. Non-volatile RAM (NVRAM) is used to store the startup configuration that the IOS reads when the router boots. some Cisco IOS features require more than the default amount of memory. and sophisticated security services usually require substantial computation). and that known vulnerabilities in other systems are of little help to the router attacker. better understood. 2. Finally. one can potentially add more network connections to a machine designed for that purpose. Thus. access speed.4. So. and ROM (PROM. a special purpose device will probably be a lower cost solution for a given level of functionality. EEPROM).Router Security Configuration Guide 2. Second. First. routing protocols can advertising selected network routes to neighbors. There are typically a number of types of memory in a router possibly including: RAM. Using filtering.4. concerns performance: a special purpose router can have much higher performance than a general purpose computer with routing functionality tacked onto it. but hardware other than Cisco would run similar software. although it may have other kinds of removable storage such as Flash memory cards. Why Have a Special Purpose Router? What are some of the motivations for using a dedicated. and other factors. because it can be designed to support more interface card slots. The mix of types and the amount of each type are determined on the basis of: volatility. thus helping protect parts of your network. it has a central processor unit (CPU). Typically. like Cisco’s Internetwork Operating System (IOS) can be smaller. purpose-built router rather than a general purpose machine with a “standard” operating system (OS)? What justifies this expense.) This means that it is potentially less vulnerable. and what justifies the bother of learning yet another system? The answer. (For brevity. in general. specialized routing software enables a fuller and more robust implementation of filtering. Flash. or CD-ROM drive. and connections to other devices. consolidating network routing and related functions on a dedicated devices restricts access to and limits the exposure of those critical functions.g. similar to any other computer. in part. a router does not have a hard disk. Filtering is useful as a “firewall” technique. the mere fact that it is different means that an attacker has one more thing to learn. Conceptual Model for Typical Router Hardware A router is essentially just another computer. Also. cost. the term IOS will be used in this guide to refer the router’s operating system and associated software.1.

fast serial. it usually does not touch the CPU. routing. most routers have an auxiliary port. and manages the operation of the interfaces. Each interface is named and numbered. the packets travel across the routing fabric from the incoming interface to the appropriate destination interface. and control. Network n Interface 0 Interface 1 . Interface cards fit into slots in a router. fast Ethernet. Interface types include Ethernet. gigibit-Ethernet. logging.1c 25 . In addition to a number of interfaces. instead. Also. HSSI. and if there is enough flash it may store more than one version of IOS. This can be important. Network 0 Network 1 . low-speed serial.] Version 1. also called the central processor or route processor. FDDI. which is frequently used for connecting a modem for router management. etc. almost all routers have a console port providing an asynchronous serial connection (RS-232).. [Do not confuse the hardware serial ports with the concept of network protocol port numbers (e.g. Figure 2-5 shows a simple representation of a notional router’s hardware structure. ISDN BRI.. It loads the configuration of the router during boot-up. When traffic is being forwarded from one network to another. token ring.. port 23 for Telnet). because the bandwidth of the channel to the CPU may be far smaller than the bandwidth of the interfaces.Background and Review memory stores the IOS (or other router OS). Only management and control traffic for the router travel to or from the CPU. and an external cable of the appropriate type is connected to the card. provides high-level services for management. Routing Fabric Interface n CPU Configuration Router Console Figure 2-5: A Notional Router’s Hardware Structure Interfaces provide the physical connections from a router to networks.. The CPU.

Description of Typical Router Software Similar to any other computer. All routers support a variety of commands that display information about what processes are running and what resources. It is executed when the router boots. they run Cisco’s Internetwork Operating System (IOS). An operational router will have a large number of processes executing to support the services and protocols that the router must support. 26 Version 1. and to avoid giving attackers additional potential avenues for attack. Each router should have a unique name to identify it.1c . If changes to a configuration are written to the startup configuration. Unneeded services and facilities should be disabled to avoid wasting CPU and memory resources. Changes made to a running configuration are usually made only in RAM and generally take effect immediately. These kinds of considerations are discussed in more detail later in this guide. a router will run a control program or operating system (OS). they are consuming. Basic security settings should be established on any router before it is connected to an operational network. It is the IOS that interprets the access control list (ACL) settings and other commands to the router.3. The startup or backup configuration is stored in NVRAM. As part of the boot process a copy of this configuration is loaded into RAM. such as CPU time and memory. and each interface should have unique network addresses associated with it.Router Security Configuration Guide 2. In the case of Cisco routers. Each router vendor supplies their own router OS.4. then they will also take effect on reboot. Changes made only to the running configuration will be lost upon reboot.

2. UDP – the User Datagram Protocol (UDP) is a connectionless. For more details see Section 4. “Routing and Routing Protocols”.5. When we speak of TCP/IP we are usually implicitly including UDP. reliable protocol. 2.4. ARP – Hosts use the Address Resolution Protocol (ARP) to acquire the MAC address of other hosts. fast-converging routing protocol.1. ICMP – the Internet Control Message Protocol (ICMP) provides the mechanisms for hosts and routers to report network conditions and errors to other hosts and routers.5. connectionless. the ping command relies on ICMP. It is a distance vector protocol that Version 1. Physical Layer 1 As previously discussed. An area is a logical collection of routers and networks. Data Link Layer 2 The IEEE and other standards that apply at this layer have also been discussed previously.5. (For example. 2.1c 27 . best effort protocol with no guarantee of delivery or confirmation of delivery. Network Layer 3 IP – the Internet Protocol (IP) provides a specification for packet formatting and an unreliable. Transport Layer 4 TCP – the Transmission Control Protocol (TCP) is a connection-oriented.Background and Review 2. It has lower overhead than TCP. RIP – Routing Information Protocol is a dynamic routing protocol that allows routers to share network information with each other. best effort delivery of those packets. Before transmitting data a connection must be established and after data transmission is complete the connection must be closed. 2.5. The protocols are grouped according the OSI layer to which they correspond.) OSPF – Open Shortest Path First is a relatively complex.2. It is an interior gateway protocol that uses a link state routing algorithm and requires that a hierarchy of areas be designed.3.5.4. Review of Router-Relevant Protocols and Layers The following sections are not inclusive of all protocols that might be of interest but are representative. the physical layer is defined by IEEE standards or similar standards that define what are primarily physical and electrical characteristics.

DNS – Domain Name System (both TCP and UDP) performs naming resolution service by translating host names into IP addresses and vice versa. NTP – the Network Time Protocol (UDP) is the Internet standard protocol for synchronizing time between network hosts and authoritative time sources. SMTP – Simple Mail Transport Protocol (TCP) is the Internet standard protocol for transmitting e-mail messages. 2. 28 Version 1. Telnet – (TCP) Enables terminal oriented processes to communicate. it is used for remote login. It is used as an interior gateway protocol. Presentation Layer 6. TFTP – the Trivial File Transfer Protocol (UDP) provides file transfers without any authentication or security.1c . HTTP – the Hypertext Transfer Protocol (TCP) is used for retrieving web pages and many related tasks. SNMP – Simple Network Management Protocol (UDP) enables a management station to trap certain information messages from network devices.5. FTP – File Transfer Protocol (TCP) enables transfers of files between hosts.Router Security Configuration Guide allows routers to only share information with their nearest neighbors.5. and Application Layer 7 These protocols are labeled (TCP) or (UDP) depending on which layer 4 protocol they are based upon. SSH – Secure Shell (TCP) provides cryptographic security for remote login sessions and other stream-oriented protocols. Session Layer 5.

6. rerouting. Attack techniques include: password guessing. DDoS attacks use a number of compromised sites to flood a target site with sufficient traffic or service requests to render it useless to legitimate users. routing protocol attacks. sequence number prediction and alteration. Session replay attacks use a sequence of packets or application commands that can be recorded. IP fragmentation attacks – to bypass filtering. Session hijacking may occur if an attacker can insert falsified IP packets after session establishment via IP spoofing. and circular redirect – for denial of service. Quick “Review” of Attacks on Routers General threats include but are not limited to: unauthorized access. Masquerade attacks occur when an attacker manipulates IP packets to falsify IP addresses. Target resources may be bandwidth.1c 29 . eavesdropping. redirect (address) attacks. masquerading. An enumeration of steps to take to improve router security. In addition to threats to a router from the network. by blocking spoofed source addresses. Resource starvation attacks usually involve flooding the router with traffic or requests designed to consume all of some limited resource. These kinds of attacks are sometimes called “route injection” attacks. and information theft. Denial of Service (DoS).Background and Review 2. Careful router configuration can help prevent a (compromised) site from being used as part of a Distributed Denial of Service (DDoS) attack. and an explanation of the tradeoffs involved is the substance of later sections of this document. possibly manipulated. SNMP attacks. and then replayed to cause an unauthorized action or gain access. or other methods. memory. or even computation. Masquerades can be used to gain unauthorized access or to inject bogus data into a network. dial up access to a router exposes it to further threats. session hijacking. Rerouting attacks can include manipulating router updates to cause traffic to flow to unauthorized destinations. Version 1.

McGraw-Hill. [9] Stevens. well written. References 2. routers. and TCP/IP. K. [3] Chappell. 30 Version 1. For the network administrator who already has basic familiarity with Cisco IOS. 1998. [5] Perlman. [6] Sacket. Managing IP Networks with Cisco Routers. L. 1997. Cisco Security Architectures. Introduction to Cisco Router Configuration.7. 1999. 1999. This book offers good explanations of all the underlying concepts. this book provides detailed information about a wide variety of topics and features. McGraw-Hill.. 1998. A. this book is a good source. Prentice-Hall. O’Reilly Associates. A practical introduction to the concepts and practices for using Cisco routers..M. with explanations of many of the concepts. (ed. Interconnections: Bridges and Routers. Prentice-Hall. but it also provides a great deal of technical background information. If you need more introductory information than this section provides. R.7. This book is primarily oriented toward network application programmers. Cisco IOS Essentials. good background reading. McGraw-Hill. an excellent source for understanding all the concepts behind networks. this book provides concepts and practical advice for using a router securely. A “classic”. A good book for learning the basics.R. For administrators already comfortable with basic operation of a router. [7] Held. Books [1] Albritton. G. 2nd edition. 1998.. 1992.1c . Cisco Press. L. W. with an emphasis on Cisco IOS. G. An excellent introduction to basic IOS operations. with no vendor emphasis.1. [2] Ballew. [8] Tannenbaum. Cisco Press. and Hundley.) Advanced Cisco Router Configuration. 1999.. J. S.. Computer Networks. [4] Chappell.Router Security Configuration Guide 2. Cisco Router Handbook.. Unix Network Programming. McGraw-Hill. 1999. This thick book provides a lot of detail on the architecture of Cisco routers and their operational concepts.

ietf. F.. W. C.2. Cisco “User Datagram Protocol (UDP)” 1993.. and Parker A series of introductory-level papers by Cisco.3. [17] Postel.. Available at: http://www.. 1996. Cisco Systems. As the official documents of the Internet Engineering Task Force. 1997. Cisco Systems. Cisco [19] Baker. RFC 1180. 2nd Edition. T. includes coverage of all the topics discussed in this section. 1999. RFC 791. Addison-Wesley. 1999. 1981. [18] Postel. [11] Cisco IOS 12. For really deep. 1997. RFC 768. Papers [13] “Internetworking Technology Overview”. with a great deal of background information. “Transmission Control Protocol (TCP)”. Cisco Product Overview. J. [16] Postel. RFC 793. 1994. and how routing works in the Internet. Available at: http://www.).R.. “Internet User’s Glossary”. 1991. J. protocols. Version 1. 1980. this book is the best source.0 Configuration Fundamentals.. [14] “OSI: The Network Layer”. A deep and detailed textbook about IP routing technologies. Routing in the Internet. “Requirements for IP Version 4 Routers”. too. (ed. This book provides a valuable reference for all the basic operation and configuration features. bit-by-bit analysis of the TCP/IP protocols. RFC 1812. J. RFC 1392. Cisco Systems Brochure.Background and Review [10] Stevens. TCP/IP Illustrated – Volume 1. [20] Socolofsky.html 2. these are the definitive sources for information about the protocols and architecture of the Internet. RFCs RFC stands for Request for Comments. Prentice-Hall. [21] Malkin. 2. 1981.html [15] “TCP/IP”. “A TCP/IP Tutorial”..7.1c 31 . technical.html . and Kale. [12] Huitema. The Protocols. All RFCs may be downloaded from http://www. Available at: http://www. C. “Internet Protocol (IP)”.7.

and Yu. J.1c . RFC 1518. Li. “Classless Inter-Domain Routing (CIDR): an Address Assignment and Aggregation Strategy”. Y. [23] Fuller.. Varadhan K.. T.Router Security Configuration Guide [22] Rekhter. 1993.. 1993. and Li. “An Architecture of IP Address Allocation with CIDR”. RFC 1519... V. T. 32 Version 1.

The room that contains the router should be free of electrostatic or magnetic interference. protecting a network with a router. Router Security Principles and Goals Routers can play a role in securing networks. If continuous operation of the router is critical.1. Protecting the Router Itself 3. Threats to the data plane usually concern violations of network security for the networks that the router supports.Router Security Principles and Goals 3. The control plane covers monitoring.1c 33 . and generally the dynamic operation of the router. configuration commands software updates authentication configuration backups routing information cryptographic negotiations audit logs network data network data Figure 3-1: Three-Plane Conceptual Model for a Router To secure a router. and generally the persistent state of the router. 3. It should have controls for temperature and humidity. a router operates on three distinct domains or planes. we must consider the possible threats to each plane. The management plane handles administration. The data or forwarding plane handles the packets transiting the router among the networks it serves. This section describes general principles for protecting a router itself.1. Conceptually. The last part of this section gives some guidance about router security policy. and managing a router securely. configuration. route table updates.1. Physical Security There are a number of ways to provide physical security for a router. Threats to the management and control planes mostly concern unauthorized access to the router. an uninterruptible power supply (UPS) should be installed and spare Version 1. or interference with router operation.

Operating System The operating system for the router is a crucial component. One should use the latest stable release of the operating system that meets the feature requirements.2 discusses management of updates to the router configuration.1.3.3. modems) used to connect to the router require storage protection. PC cards. 34 Version 1. This is a false savings.g. Decide what features the network needs.. for example. To help protect against some denial of service attacks.1c . In the Internet Service Provider community. Section 3. and to allow it to support the widest range of security services. Configuration Hardening A router is similar to many computers in that it has many services enabled by default. and Sections 4 and 8 include information on Cisco’s IOS operating system. However. Many of these services are unnecessary and may be used by an attacker for information gathering or for exploitation.Router Security Configuration Guide components kept on hand. Physical devices (e. * Also. it is considered an industry best practice to equip every operational router with as much memory as it can hold. and use the feature list to select the version of the operating system.2 discusses the management of updates to the operating system. the router should be configured with the maximum amount of memory possible. adding memory to an operational router requires taking that router out of service. 3. Section 3.3. the router should be placed in a locked room accessible only to authorized personnel.2. the very latest version of any operating system tends not to be the most reliable due to its limited exposure in a wide range of network environments. on the grounds that memory costs money and therefore a router should be purchased with the minimum amount of memory it needs to supports its task. Also. The incremental cost of extra memory is usually small compared to the total cost of a fully configured router. and the added performance and flexibility that the extra memory will provide is almost always worthwhile when amortized over the number of users and services that depend on the router for connectivity over its service lifetime. * Some readers might balk at this recommendation. Unnecessary services should be disabled in the router configuration.1. 3.

Interior routers may impose some restrictions on the traffic they forward between networks. and to protect the integrity of the routing information it uses to forward traffic. Most of the techniques described in this guide are applicable to backbone routers. Protecting the Network with the Router 3.Router Security Principles and Goals 3.1c 35 . they will usually employ an Interior Gateway Protocol to manage routes. The level of trust between the networks connected by a backbone router is usually very low. but may need to be modified or adapted to specific applications. Typically. but for this discussion we will examine three fundamental ways in which routers are employed. The traffic between the different networks that make up the Internet is directed by backbone routers. Version 1. Roles in Network Operations and Security Routers perform many different jobs in modern networks. Internal LAN 1 Internal LAN 2 Router Internal LAN 3 Figure 3-2: An Interior Router Connects an Organization’s Internal Networks Backbone Routers A backbone or exterior router is one that forwards traffic between different enterprises (sometimes called different ‘autonomous systems’).1. Backbone routers typically employ Exterior Gateway Protocols to manage routes.2. without imposing any restrictions on it.2. Most of the directions in this guide are useful for interior routers. The networks connected by an interior router often share the same security policy. backbone routers are designed and configured to forward traffic as quickly as possible. Interior Routers An interior router forwards traffic between two or more local networks within an organization or enterprise. and the level of trust between them is usually high. The primary security goals for a backbone router are to ensure that the management and operation of the router are conducted only by authorized parties. Configuring backbone routers is a very specialized task. If an enterprise has many interior routers.

The key aspect of a border router is that it forms part of the boundary between the trusted internal networks of an enterprise.g. or it may depend entirely on static routes.1c . The firewall provides additional access control over connections and network traffic. A border router may employ routing protocols. a border router is not the only component at the boundary. the border router acts as the first line of defense and is known as a screening router. 36 Version 1. It contains a static route that passes all connections intended for the protected network to the firewall. the Internet). and untrusted external networks (e. Using a firewall and a router together can offer better security than either one alone. The firewall may also perform user authentication. many enterprises also employ a firewall to enforce fine-grained security policy. Internet Router Internal Networks Figure 3-4: A Border Router Connects Internal Networks to an External Network Typically. In the Figure 3-5.Router Security Configuration Guide Router Router Router Router Figure 3-3: Backbone Routers Connect Many Networks Border Routers A border router forwards traffic between an enterprise and exterior networks. It can help to secure the perimeter of an enterprise network by enforcing restrictions on the traffic that it controls.

However. FTP file name). It also offers an intermediate area. some routers need two filters to handle bi-directional services. while others have filters that apply only in one direction. Internet Router Premises or Gateway router Router Internal or Local net router Internal Network Firewall Figure 3-6: A Two-Router Firewall Configuration for a Network Boundary All of the directions in this guide are suitable for border routers. routers typically do not have the capability to filter on the content of services (e.2. the router can enforce security policy. a user on System A telnets to System B. (Many services are bi-directional. The DMZ is often used for servers that must be accessible from the Internet or other external network. destination port.2. For example. Filters are also important for their ability to enforce addressing constraints. often called the de-militarized zone (DMZ) between the two routers. Some routers can even filter on any bit or any pattern of bits in the IP header. source port. Packet Filters for TCP/IP A packet filter for TCP/IP services provides control of the data transfer between networks based on addresses and protocols.) Most routers can filter on one or more of the following: source IP address. This configuration offers two points at which policy can be enforced. Some routers have filters that apply to network services in both inbound and outbound directions. and then another router between the firewall and the trusted internal networks. in Figure 3-1. the router should enforce the constraint that packets sent Version 1. Routers can apply filters in different ways. In that role. So.Router Security Principles and Goals Internet Router Firewall Protected Network Figure 3-5: A Simple One-Router Firewall Configuration for a Network Boundary Another approach is to position one router at the connection between the external networks.g. 3. rejecting protocols and restricting ports according to the policies of the trusted network. and protocol type. destination IP address.1c 37 . For example. and System B sends some type of response back to System A. Packet filters are especially important for routers that act as the gateway between trusted and untrusted networks.

This is a form of ingress filtering. fall back to prohibiting services that are commonly not needed.1c . as the length grows the filter becomes more complex and more difficult to troubleshoot. it is not possible to follow the strict security guideline discussed above. the length and ordering of a packet filter rule set can affect the router’s performance. Applying Packet Filters: Permit Only Required Protocols and Services Carefully consider what network services will be allowed through the router (outbound and inbound) and to the router. Two key characteristics of TCP/IP packet filters are length and ordering. (Note: This discussion is applicable to the packet filtering facilities of Cisco routers. Generally. One method of ordering involves placing those rules that will handle the bulk of the traffic as close to the beginning of the filter as possible. In cases where only certain hosts or networks need access to particular services. consult its documentation for details). If possible. Consequently. Create a set of filtering rules that permit the traffic identified on the list. add a filtering rule that permits that service but only for the specific host addresses or address ranges. You must carefully create filter rules in the proper order so that all packets are treated according to the intended security policy. Cisco filtering is discussed in detail in Section 4. For example. This guideline is especially important for border routers. The following two tables present common services to restrict because they can be used to gather information about the protected network or they have weaknesses that can be exploited against the protected network. or are known to be popular vehicles for security compromise. When the router analyzes a packet against a filter the packet is effectively compared to each filter rule in sequential order. The order of the rules in a packet filter is critical. The number of rules in a filter determines its length. Similarly. and most packet filtering firewalls.Router Security Configuration Guide from the internal or protected network (right to left) must bear a source address within a particular range. The first table lists those services that should be completely blocked by a typical border router. In that case. Make a list of the services and protocols that must cross the router. If no match is found then the packet is denied due to the implicit deny rule at the end of the filter. with each rule either accepting or denying a certain set of packets. most other kinds of routers. This is sometimes called egress filtering. A filter consists of one or more rules. Unless you have a specific operational need to 38 Version 1. the network firewall host might be the only address authorized to initiate web connections (TCP port 80) through the router. use the following guideline for creating filters: those services that are not explicitly permitted are prohibited.3. and those that the router itself needs for its operation. If you have a router made by a company other than Cisco Systems. the router should enforce the constraint that packets arriving from the Internet must bear a source address outside the range valid for the protected network. Applying Packet Filters: Reject Risky Protocols and Services Sometimes. and prohibits all other traffic. If a match is found then the packet is either permitted or denied and the rest of the filter is ignored.

Table 3-1: Services to Block Completely at a Border Router Port (Transport) 1 (TCP & UDP) 7 (TCP & UDP) 9 (TCP & UDP) 11 (TCP) 13 (TCP & UDP) 15 (TCP) 19 (TCP & UDP) 37 (TCP & UDP) 43 (TCP) 67 (UDP) 69 (UDP) 95 (TCP & UDP) 111 (TCP & UDP) 135 (TCP & UDP) 137 (TCP & UDP) 138 (TCP & UDP) 139 (TCP & UDP) 177 (UDP) 445 (TCP) 512 (TCP) 515 (TCP) 517 (UDP) 518 (UDP) 540 (TCP) 1434 (UDP) 1900. Version 1.6063 (TCP) 6667 (TCP) 12345-6 (TCP) 31337 (TCP & UDP) Service tcpmux echo discard systat daytime netstat chargen time whois bootp tftp supdup sunrpc loc-srv netbios-ns netbios-dgm netbios-ssn xdmcp netbios (ds) rexec lpr talk ntalk uucp Microsoft SQL Server Microsoft UPnP SSDP NFS X Window System IRC NetBus Back Orifice Table 3-2 lists some services on the internal network or on the router itself that should not be accessible to connections from the external networks.Router Security Principles and Goals support them. 5000 (TCP & UDP) 2049 (UDP) 6000 . the protocols listed in Table 3-1 should not be allowed across the router in either direction.1c 39 .

Various organization in the US DOD maintain such lists. In most cases filtering rules should apply both ingress and egress filtering. rdump syslog new who Standard Ports and Protocols Some organizations maintain a list of standard ports and protocols that should be allowed or supported on their networks.1c .Router Security Configuration Guide Table 3-2: Some Services to Block at the Router from External Clients Port (Transport) 79 (TCP) 161 (TCP & UDP) 162 (TCP & UDP) 513 (TCP) 513 (UDP) 514 (TCP) 514 (UDP) 550 (TCP & UDP) Service finger snmp snmp trap rlogin who rsh. 40 Version 1. rcp. and should be dropped by a border router. unroutable. or illegal address range. allowing only those ports and protocols mandated by the standard list. The principles to apply on border routers are listed below. especially on border routers. any other traffic is attempting to claim a bogus source address.) • Reject all traffic from the external networks that bears a source address belonging to the internal networks. For networks that are subject to such lists. Address Filtering Router filters should also be used to protect against IP address spoofing. Traffic bearing such spoofed addresses is often part of an attack. traffic sent from the external networks should always bear a source address from some range other than those assigned to the internal networks. • Reject all traffic from the internal networks that bears a source IP address which does not belong to the internal networks. including blocking reserved addresses. and is almost certainly erroneous or malicious in nature.) • Reject all traffic with a source or destination address belonging to any reserved. rdist. (Assuming that addresses are assigned correctly. (Legitimate traffic generated by sources on the internal networks will always bear a source address within the range or ranges assigned to the internal networks. it is best to take the first approach. and rejecting all others. and the Defense Information System Agency (DISA) is attempting to manage the creation of a standard list for the entire DOD.

Some routers and firewalls offer specialized facilities to mitigate TCP SYN flood attacks. Prevent malicious traffic from entering the common network from the enterprise network.Router Security Principles and Goals 3. Many of these attacks require use of invalid or spoofed source addresses.3.3. Invalid or otherwise untraceable source addresses are often used to hide the actual source of an attack. 3.6). There are measures that network administrators can take to protect networks from DoS attacks and lessen their effects. detection will depend on reviewing the router logs on a regular basis.3). Though DoS attacks can be viewed as tolerable annoyances.1c 41 . they can have serious consequences if they occur during a time of crisis. These measures require some cooperative effort between those who administer hosts. logging these violations can help identify compromised hosts that need to be removed from your network. as long as the resources of a network are limited and openly available they will be vulnerable to attack. You can defend against some individual DoS attacks locally by rejecting packets with invalid source addresses as they arrive at a border router (see Section 4. these measures must be planned and in place before an attack occurs. 1. First.4. Black hole routing can also be useful. you should log occurences of bad packets. Denial of Service (DoS) refers to willful attempts to cause such disruptions. and provider access.2. For example. at both border and interior routers. Where possible. Also. Access control lists are a general filtering facility available on all routers (see Section 4. Mitigating Denial of Service Attacks Loss of service or severely degraded network performance can result from a variety of causes.4. and works on all routers (see Section 4. Most Cisco routers support a facility called Unicast Reverse-Path Forwarding Verification that uses the route table to detect and drop improperlyaddressed packets (see Section 4. At the enterprise level there are three primary strategies for combatting DoS attacks. on Cisco routers this facility is called Version 1.5). There are several ways to filter out these improperly-addressed packets.2). To be effective.7). it is important for every network administrator to help reduce the number of DoS attack launch platforms. Of course. described in detail below. keep hosts secure and eliminate compromised hosts from the network immediately. 2. invalid addresses are used in SYN flood attacks to ensure that the TCP handshake on the target host times out waiting for a response (see Section 6.3. network devices. Coordinate protective measures against distributed DoS attacks with network access providers and/or backbone administrators.2). There is no complete solution to the DoS problem. router services that support attacks or attack amplification should be disabled (see Section 4. Configure and deploy local protective measures. There are several mechanisms available on routers to thwart certain kinds of DoS attacks. Do not let your network be the origin point for a DoS attack.

3. the source point access provider. Router facilities may also be supplemented by commercial anti-DoS products that provide finer-grained filtering and attack detection. In some cases.3). as the packets of an ICMP flood converge at the uplink. from the access provider. DoS mitigation necessarily requires cooperative action “upstream. or even from the administrators of the attacking hosts. 42 Version 1. Routers are a part of the solution. legitimate traffic is crowded out by bogus traffic and packets are lost to traffic flow control. they can take steps to block and trace back bad traffic as the attack ramps up.e.” i.Router Security Configuration Guide TCP Intercept (see Section 4. Connections and data transfers are starved and eventually time out or hang because they are unable to resynchronize. router traffic rate control or quality of service facilities can be used to protect critical services from the full effects of DoS attacks (see Section 4. A border router cannot control the type or overall volume of traffic that is sent to it. (possibly from) the transport provider.1c . but cautious design. If no such quality of service monitoring exists. and cooperation among network administrators are also necessary. It is still prudent to be prepared to handle well-known DoS attacks using facilities already available. If your access provider performs statistical monitoring of traffic.6). and certainly there will be novel kinds of DoS attacks discovered in the future. For example.3. There is no set of methods that can completely counter all known DoS attacks. contingency planning. then the network being attacked will need to actively request its access provider filter out offending traffic.

1c 43 . 1. The management network should include only identified administration hosts and a spare interface on each router. For example. Define clearly the capabilities of each level or role in the router security policy. Establish a dedicated management network.1. There are two types of access: local and remote. or more than one administrative role. If an attacker can collect network traffic during remote access then he can capture passwords or community strings. LAN 1 Internet Router Firewall Router LAN 2 Management LAN Administration Host Logging Host Figure 3-7: Using a Management LAN for Administration 2. Local access usually involves a direct connection to a console port on the router with a dumb terminal or a laptop computer.3 shows how to set up a Cisco router to support SSH encryption. However. (Section 5. Remote access typically involves allowing telnet or SNMP connections to the router from some computer on the same subnet or a different subnet. Section 5. one role might be “network manager”. Another method is to encrypt all traffic between the administrator’s computer and the router. Managing the Router 3. administrators authorized to assume that role Version 1. and administrators authorized to assume that role may be able to view and modify the configuration settings and interface parameters. Another role might be “operators”. there are some options if remote access is required.3. packet filters can be configured to permit only the identified administration hosts management access to the router. It is recommended to only allow local access because during remote access all telnet passwords or SNMP community strings are sent in the clear to the router.Router Security Principles and Goals 3.) In either case.3.2 shows an example of setting up IPSec encryption with a Cisco router and Windows 2000. Figure 3-7 shows an example of this. Access Mechanisms for Administrators Controlling access to a router by administrators is an important issue. In addition to how administrators access the router. there may be a need to have more than one level of administrator.

The logging level may need to be modified based on how much of the log information is useful. If the tests are successful then restore or reconnect the interfaces on the router. direct the router to at least two different reliable time servers (via NTP) to ensure accurate and reliable 44 Version 1. All routers are capable of maintaining their own time-of-day. If the tests are not successful then back out the update. Logging Logging a router’s activities and status offers several benefits. the administrator can tell whether the router is working properly or whether it has been compromised. Set the level of logging on the router to meet the needs of your security policy. 3. to perform the update. Send the router logs to a designated log host. Instead. which is a separate computer whose only job is to accept and store logs.1c . Configuring logging on the router should be done carefully.3. By checking over the logs periodically. it is best to keep the number of fully privileged administrators to a minimum. and if necessary install additional memory. Accurate timestamps are important to logging. Load the update for either the operating system or for the configuration file. Determine the memory required for the update. Set up and test file transfer capability between the administrator’s host and the router. These updates are necessary for one or more of the following reasons: to fix known security vulnerabilities. the administrator should complete the following checks. The log host should be connected to a trusted or protected network. Harden the log host by removing all unnecessary services and accounts. usually after regular business hours. A sound understanding of normal operation and its reflection in the logs will help you to identify abnormal or attack conditions. Two areas that should be logged are (1) matches to filter rules that deny access. it can show what types of probes or attacks are being attempted against the router or the protected network. and expect to modify the log settings as the network evolves. Back up the current operating system and the current configuration file to the administrator’s computer. or an isolated and dedicated router interface.3. In some cases.3. Updating the Router Periodically the router will require updates to be loaded for either the operating system or the configuration file. The most important thing to remember about logging is that logs must be reviewed regularly.2. to improve performance or support new features (perhaps some that allow more advanced security policies). Perform tests to confirm that the update works properly. Using the information in a log. and (2) changes to the router configuration. Shut down or disconnect the interfaces on the router. In general. the administrator should follow procedures similar to the following. Before updating. 3. After obtaining an update from the router vendor (and verifying its integrity). you can gain a feeling for the normal behavior of your network.Router Security Configuration Guide might be authorized only to clear connections and counters. but this is usually not sufficient. Schedule the required router and network downtime.

This will allow you to trace network attacks more credibly. Finally. compromise of the log host).Router Security Principles and Goals of time information. write down the procedures. Another important aspect of lifetime security is preparing for problems. Direct the logging host to reliable time servers. so that it can be used in a forensic investigation or even prosecution.g. Include the steps for capturing the compromised state of a router in your recovery plan.1c 45 . it is highly desirable to preserve the evidence. testing. Keeping up to date backups of router configurations and installed IOS releases is essential for quick and reliable recovery from security compromises or simple hardware failures. 3. consider also sending the logs to write-once media or a dedicated printer to deal with worst case scenarios (e. and then exercise the plan periodically so that all the participants understand their roles. In the case of a security compromise. Your recovery plan must be coordinated with your security policy (see next section). Version 1. Include a timestamp in each log message.3. Plan your recovery actions. Operational Security Management Maintaining the security of a router over its operational lifetime requires regular assessment. and correction.4.

and their security is a vital part of the overall security for the networks they serve. Security policy usually includes strict rules about access to this layer. usually called ‘Console’ or ‘Control’ ports. Router security policy should define rules for where and how these ports may be used. The next outermost zone of the diagram is the dynamic configuration of the router. The next innermost zone of the diagram is the stored software and configuration state of the router itself.1. Security Policy for Routers Routers are an important part of a network. below. the user names and passwords. and the access controls for direct access to the router’s command interface. are also very important. and management of the router satisfy your security policy? 3. in terms of both administrative roles and network mechanisms. then he will also gain control of the outer two layers. Other pieces of dynamic information. Figure 3-8: Layered View of Router Security The innermost zone is the physical security of the router. configuration.4. such as interface status. If an attacker can compromise the dynamic configuration of a 46 Version 1. The security of each layer depends on the security of the layers inside it. A Conceptual Basis for Router Security Policy Figure 3-8. Router Security Layers Physical Integrity of the Router Core Static Configuration of the Router Dynamic Configuration and Status of the Router Network Traffic through the Router Corresponding Access • Physical access • Electrical access • Administrative access • Software updates • Routing protocols • Access to the network that the router serves. shows a layered view of the security of a router. What does it mean for a router to be secure? One simple way to define the security of a router is this: does the operation. The route tables themselves are the most obvious part of this. these ports usually provide special mechanisms for controlling the router. Some important aspects of the stored configuration are the interface addresses.4. physical access must be controlled to provide a solid foundation for the overall security of the router. and audit logs. therefore. Any router can be compromised by an attacker with full physical access. Most routers offer one or more direct connections. If an attacker can compromise either of these. particularly the stored configuration.Router Security Configuration Guide 3.1c . ARP tables.

1c 47 . not particular commands or mechanisms – When the policy specifies the security results to be achieved. The high-level requirements of the network security policy must be reflected in the configuration of the router. he can compromise the outermost layer as well.4. • Specify policy for all the zones identified in the figure above – Begin with physical security. The router policy might specify the particular rules to be enforced by the router to prevent unauthorized remote administration. The operator. 3. the network that a router serves will have a security policy. access mechanisms. and for traffic flow. In other regards. The router security policy might include only two: administrator and operator. a network security policy might define three roles: administrator. permissions. For example. the policy is more portable across router software versions and between different kinds of routers. 3. The overall network security policy may include rules about this. Each of the roles would be granted privileges in the router policy that permit them to fulfill their responsibilities as outlined in the network policy. the network security policy might forbid administration of the router from anywhere but the local LAN. and administrative roles. operator. and user. the router policy will involve far more detail than the network policy. defining roles. The roles defined in the router security policy will usually be a subset of those in the network policy. In some cases.4. the dynamic configuration. Version 1. Security policy for a router should include rules about access to this layer. the router enforces network policy. The outer zone of the diagram represents the intra-network and inter-network traffic that the router manages. rules of conduct. identifying permitted protocols and services.2. For example. and probably in the router security policy.3. and the router policy must reflect this. Router Security Policy and Overall Network Security Policy Typically. although it is sometimes overlooked. and responsibilities. Creating a Security Policy for a Router There are several important tips to remember when creating the security policy for a router: • Specify security objectives. The policy for a router must fit into the overall framework. The router security policy might grant the operator login privileges to the router so that they can access the router logs. and work outwards to security for the static configuration. The rules of conduct for administering the router should clarify the application of the network rules to the router. rather than a particular command or mechanism. for example.Router Security Principles and Goals router. might be held responsible by the network security policy for periodic review of the audit logs.

due to performance concerns or differences in the security policies of the different networks served. In some cases. notify all individuals authorized to administer the router and all individuals authorized for physical access to it. When drafting a policy. avoid specifying the particular mechanisms in the policy. Router Security Policy Checklist The checklist below is designed as an aid for creating router security policy. 3. A security policy must be a living document. a new firewall) • Detection of an attack or serious compromise When the router security policy undergoes a revision. Carefully check your router’s security policy against any applicable high-level policy. Update the router policy to reflect changes in the network policy. explicitly permit those. the policy should clearly state any limitations or restrictions that can be enforced. keep most of the directives and objectives high-level. In particular.g. • New connections made between the local network and outside networks • Major changes to administrative practices. or whenever the security objectives for the router change. Maintaining policy awareness is crucial for policy compliance. In these kinds of cases. After drafting a policy.1c .g.4. Make it part of the security practices of your organization to regularly review the network security policy and the router security policy. some organizations have high-level policies that impose specific requirements on the contents of individual network security policies. a new VPN) or new network components (e. procedures. It may be necessary to revise the router security policy whenever there is a major change in the network architecture or organizational structure of network administration.Router Security Configuration Guide • Services and protocols that are not explicitly permitted should be denied – When representing the network policy in the router policy. Finally. concentrate on services and protocols that have been identified as explicitly needed for network operation. it may not be practical to identify and list all the services and protocols that the router will explicitly permit. examine the router security policy and revise it as needed whenever any of the following events occur. step through the list and check that each item is addressed in your policy. or staff • Major changes to the overall network security policy • Deployment of substantial new capabilities (e. A backbone router that must route traffic to many other networks cannot always enforce highly tailored policies on the traffic flowing through it. 48 Version 1. and deny everything else.4. to ensure that it meets all the requirements.

Router Security Principles and Goals Physical Security • • Designates who is authorized to install. de-install. Defines the recovery procedures and identifies individuals responsible for recovery. Defines the key management policy for long-term cryptographic keys (if any). Defines the audit log policy for the router. and move the router. including outlining log management practices and procedures and log review responsibilities. • • • • • • • • • • • Dynamic Configuration Security • Identifies the dynamic configuration services permitted on the router. log book. Defines the management policy and update intervals for long-term secrets. Designates procedures and limits on use of automated remote management and monitoring facilities (e. or evidence of tampering with the router. Designates who is authorized to make physical connections to the router. TACACS+. Defines controls on placement and use of console and other direct access port connections. Defines recovery procedures for the event of physical damage to the router. Designates who is authorized to perform hardware maintenance and to change the physical configuration of the router.g lifetime. Include a list of conditions that require passwords to be changed (e.g.1c 49 . staff changes. NTP. and networks permitted for logging in to the router remotely. and the networks permitted to access those services. and SNMP. change recording. Version 1. Designates protocols.g. Defines procedures and practices for making changes to the router static configuration (e. in the case of compromise of the router’s static configuration. review procedures) Defines the password policy for user/login passwords. • • • Static Configuration Security • Designates who is authorized to log in directly to the router via the console or other direct access port connections. Designates who is authorized to assume administrative privileges on the router. RADIUS. and for administrative or privilege passwords. compromise) Designates who is authorized to log in to the router remotely. procedures. such as those for routing protocols. SNMP) Outlines response procedures or guidelines for detection of an attack against the router itself.

• • 50 Version 1. • Compromise Response • Enumerates individuals or organizations to be notified in the event of a network compromise. NTP).g. and services to be permitted or filtered by the router. including provision for preserving evidence and for notification of law enforcement. for each interface or connection (e. manual setting. authorities. inbound and outbound). Identifies key agreement and cryptographic algorithms authorized for use in establishing VPN tunnels with other networks (if any).1c . and identifies procedures and authorities for authorizing them. Describes security procedures and roles for interactions with external service providers and maintenance technicians. Defines response procedures. and objectives for response after a successful attack against the network. Designates mechanisms and policies for setting or automating maintenance of the router’s clock (e. ports.g.Router Security Configuration Guide • Identifies the routing protocols to be used. • • Network Service Security • Enumerates protocols. Identifies relevant configuration information to be captured and retained. and the security features to be employed on each.

[3] Stevens. 1994.Router Security Principles and Goals 3. 2000.. and the Dynamic and/or Private 2002. O’Reilly & Associates.5. Building Internet Firewalls. T.B. W. Cooper.iana. the last of these documents was RFC 1700. References 3... Web Sites and On-Line Resources [5] Cisco Internetworking Technology Overview http://www.5. [6] IANA Port and Protocol Number Assignments http://www. The ‘port-numbers’ file contains the listing of all registered port numbers. in addition to its Cisco-specific material. The overview “Routing Basics” is a fine introduction to IP routing. 1999. Hardening Cisco Routers.1c 51 .5.2. constantly updated and revised when new information is available and new assignments are made. includes a good section on physical security. 3.iana. It is also included on every Cisco documentation CD. 2nd Edition. Version 1. [4] Akin. O’Reilly & Associates. K. these numbers were documented through the RFC document series. [2] Held. S. with extensive updates for newer technologies. Cisco Security Architectures.1. E. IANA houses the many unique parameters and protocol values necessary for the operation of the Internet and its future development. Types of numbers range from unique port assignments to the registration of character sets. Addison-Wesley. In the past. This revised edition includes all the sound background of the original. A pragmatic and detailed guide to securing Cisco routers..R.D. great technical background for any network analyst. This book includes excellent general advice about router and router-related network security. Books and Manuals [1] This site contains a set of excellent technical overviews for a wide variety of networking technologies. and Zwicky. A seminal overview of network boundary security concerns and techniques.. and Hundley.. Since that time. The port numbers are divided into three ranges: the Well Known Ports. McGraw-Hill. the Registered Ports. Volume 1. which is also now outdated. TCP/IP Illustrated. D. The most comprehensive and readable guide to the TCP/IP protocol suite. the assignments have been listed in this directory as living http://www.

Cisco Systems. are available under the SAFE web page: http://www.html A complex and highly detailed architecture and practices document for setting up enterprise networks. 52 Version 1. and presents a network architecture designed to protect against it. K. SANS.Router Security Configuration Guide [7] The RFC Editor Site http://www. and related ones. Security Consensus Operational Readiness Evaluation. This RFC provides extensive information about network and system security policies.) “Site Security Handbook”.rfc-editor. A detailed checklist of security and operational conditions to check for in the audit of a router. September 1997.sans. [11] [8] “Network Security Policy: Best Practices White Paper”. especially for systems connected to the Internet. available under: http://www. RFC 2196. (ed. This is the main site for looking up Internet RFCs.1c . 2000. [10] “Cisco SAFE: A Security Blueprint for Enterprise Networks”.com/warp/public/126/secpol. 2000. Cisco Systems. Cisco White Paper. “Cisco Checklist” B. The retrieval service supports a variety of keyword searches. This white paper. Cisco White Paper. [9] Naidu. as well as straight by-number lookup. This detailed white paper describes a threat model for enterprise networks.

Router Security Principles and Goals Version 1.1c 53 .

20/16 14.0. it is not a design for a secure network.18/24 eth 0 East net access eth 1 14.2.2/16 Facility Network 14.10.20 Internet 7.10.0/24 User Host 14.64/24 Protected Enclave 14.1.20 Remote Remote Network 7.250/24 Telephone Network Public Web Server 14.2.2/24 Figure 4-1: Example Network Architecture Figure 4-1 is simply a vehicle for presenting security guidance about routers.9.2.6. 5.2.1/24 Mail Server and 6.3/24 eth 0/1 (firewall) modem .2. The structures and addresses illustrated here are used for all of the examples in Sections 4.9.0/24 User Host 14. However. 54 Version 1. this architecture reasonably reflects the kinds of networks found in many organizations.64/24 Remote Host South User Host 14.6/24 FTP & Web Server North (Perimeter router) eth 0/1 Authentication Server 14.6/24 Admin Server eth 0/0 14.250/16 Central eth 0/1 14.0/24 eth 0/0 14.2/24 LAN 2 14.Router Security Configuration Guide 4.1. Implementing Security on Cisco Routers The diagram below shows a simple network configuration.9.6/24 14.250/24 DNS Server eth 0/0 modem LAN 1

use encryption to protect the confidentiality and integrity of the remote administration connection (see Sections 5. only slowed down. or they may find ways to circumvent the physical security precautions. There are things that can be done to make this more difficult. “Step 1 Configure the router to boot up without reading the configuration memory (NVRAM). especially routers and switches. The details of the procedure varies between router models. should be located in a limited access area. 4. If possible. This area should be under some sort of supervision 24 hours a day and 7 days a week. and configuration issues. are especially important because they are often used to protect segments of the network and can also be used for launching attacks against other network segments. or electronic monitoring. Network infrastructure components. an individual with physical access can gain full privileged (enable) access to a Cisco router without using a password. where applicable. Use access lists to limit remote administration access to hosts that enjoy reasonable physical security. This is sometimes called the test system mode. physical security mechanisms and policies must not make access too difficult for authorized personnel. In practice. To illustrate one reason why physical security is critical to overall router security. remote administration concerns. This can be accomplished through the use of guards.Advanced Security Services 4. An administrator (or an attacker) can simply connect a terminal or computer to the console port and follow the procedure below (taken from “Password Recovery Process” in [1]). It is always a matter of time and effort. then consider ways of protecting the machines used for remote administration.3). system personnel. If remote administration is used to configure and control routers. Version 1. One of the best additions to the security features of a computer network is to limit access. These include physical access. this area should only be accessible by personnel with administrative responsibilities for the router. but a knowledgeable attacker with access can never be completely defeated. as well as for your routers. consider the password recovery procedure for Cisco routers. Router Access Security This section discusses the various mechanisms used to protect the router itself.1. Using this procedure. software protection.2 and 5. If possible. This problem is not only confined to network devices but is also true of computers and any other electrical or mechanical device. Network equipment.1. user account protection. Physical Security Once an individual has physical access to a piece of networking equipment there is no way to stop him from modifying the system.1c 55 . but always includes the following basic steps. like routers.1. and the networks they use to communicate with the router. When thinking about the security of your network it is important to consider these issues for all your systems.

Step 3 Access enable mode (which can be done without a password if you are in test system mode).1.” Anyone with experience or training using Cisco routers can parley physical access into full privileged administrative access. Some administrators connect a modem to the auxiliary port to facilitate remote administration via dial-up. Step 5 Reconfigure the router to boot up and read the NVRAM as it normally does. The rest of this section describes configuration steps that you should take to lock down your routers.3). If at all possible. Failure to do so will usually result in the router coming up in an insecure state on subsequent reboots. The console (con) and auxiliary (aux) ports on Cisco routers are used for serial connections to the router. An operational security concern closely related to physical security is physical operating environment.) A second reason for controlling physical access to the router involves flash memory cards.Router Security Configuration Guide Step 2 Reboot the system. The area should also be controlled for temperature and humidity. If done carefully and well. or replace an old one. 56 Version 1. because a short power outage can leave some network equipment in undetermined states. all routers should be placed on an Uninterruptible Power Supply (UPS). In many cases. the auxiliary port should be disabled (see Section 4. Step 4 View or change the password. if you need to use the password recovery procedure for any reason. Step 6 Reboot the system. They could then boot the router with their flash.1c . some of the smallest models have only a console port. or erase the configuration. do not neglect to restore the system boot settings after regaining access to the router. routers are sensitive to extreme temperature and humidity. In general. Most Cisco routers have both a console and an auxiliary port. Many Cisco router models offer PC-Card slots or CompactFlash slots that can hold additional flash memory. Permitting direct dial-in to any vital piece of network infrastructure is potentially very risky. (Note: Step 5 is very important. The primary difference between the two ports is that the password recovery mechanism can be used on the console port only. It is important to apply these measures before connecting a new router to any potentially hostile network. the procedure takes only a couple of minutes. Like most networking equipment. If a router is not located in an environmentally friendly area then it may operate in unexpected ways and degrade its security. A room where routers are located should be free of electrostatic and magnetic interference. An attacker with physical access to a router on your network can install a flash memory card. Routers equipped with these kinds of slots will give preference to memory installed in a slot over memory installed in the chassis. and should be set up only when timely access by other means is not feasible. the auxiliary port is unused. this kind of attack can be very difficult to detect. This is also a personnel safety issue. thus causing the router to run their IOS version and configuration. The best defense against it is good physical security.

Version 1. etc. Cisco IOS 11. There are several configuration modes on a Cisco router.2. The recommended minimum IOS release is IOS 12. and logging router functions. including the ability to change the configuration.0. Router Configuration and Commands (IOS) After connecting to a router and initially logging in. • access-list: There are two types of IP named access lists.6 for a brief description and Section 4. To enter the global configuration mode (config) type the command configure terminal .1 was an early version of IOS Release 12. In the global configuration mode a wide variety of overall router features and settings can be changed: banners. lines. Router Software Versions Cisco issues new IOS versions and upgrades fairly frequently. auxiliary port and virtual terminal lines. 4. Cisco IOS releases. making it a significant administrative burden to keep all the routers on a large network up to date. For more details on IOS upgrades. or Vlan2. A second problem is that the early versions of new IOS releases can be less robust than more mature. authorization. • line (config-line) is used to set up the console port. log in and enter the command show version. The recommended newest release would be the most recent “GD” version that is at least a month old (at the time of this writing. the so-called ‘new model’. routing protocols. The list below describes some of the sub-modes. 12. while 12. but not cutting-edge.5 and 8. EXEC mode gives limited access to the command set of the router. see Sections 4. To check your IOS version.1c 57 .9 was a mature version of Release 12). Keep your IOS as up to date as is practical. extended (config-ext-n) and standard (config-std-n). and add new security features. authorization. Typing the enable command at an EXEC mode prompt will give access to the privileged EXEC mode.0.e. the system is in user mode also known as EXEC mode. commonly abbreviated “config t”. logging.Advanced Security Services 4. Access-list mode is used for building named access lists. which can be used instead of numbered lists. and much more. There are sub-modes which are used to configure specific settings for interfaces. is reserved for the privileged EXEC mode.21).1.1 and later offer a comprehensive model for authentication. Privileged EXEC mode is sometimes called ‘enable mode’. See Section 4. later versions (i. 12.6 for more details.1.0.1. Newer versions of IOS fix bugs and vulnerabilities that existed in the older versions. and accounting (AAA). Access to all the router commands. A good approach to this problem is to maintain operational routers with recent.3. access lists. Ethernet 0/1. • interface (config-if) is used to configure aspects of a particular interface like FastEthernet0. In addition to the standard authentication. • route (config-route) is where specific parameters can be set and modified for a selected routing protocol.1.3. routing protocols. authentication systems. This will allow others to find the bugs in the newer versions (and get them fixed).

Note that the loopback address cannot be the same as the address of any other interface. but routers also generate some network traffic. For a border router. and TFTP. that connection must have some source address. Commands to set source interface bindings are given with the discussion of each service in the rest of the guide. When a router is configured to use the loopback interface for services. to define one loopback interface. SNMP. NTP. not the external network. 4. typically a router will select a source address from one of the addresses bound to one of its network interfaces. Logins. Router Network Traffic and the Loopback Interface The pimary job of a router is forwarding traffic between networks. the loopback’s address usually should be in the range of the internal or DMZ network.11. When the router initiates a network connection. the loopback interface’s address does not appear in any route-based network maps. End with CNTL/Z. consult [5]. in configuring Cisco routers. a statement about 58 Version 1. a statement that all use of the device must be authorized by the owning organization. It is considered best practice. and Accounts Logins and Banners A login banner. Privileges.4. This can be problematic in several ways.2. A legal notice usually includes a ‘no trespassing’ warning. nor can it be part of the same network as any other interface.1.250 255.255. one per line. should be set up on each operational router. Passwords.5.Router Security Configuration Guide 4. which includes a legal notice.1. and designate it as the source interface for most traffic generated by the router itself. Central# config t Enter configuration commands.1c . Adopting this practice yields several benefits for the overall stability and security management of a network. it is possible to configure the security of other devices in the network more tightly. For further discussion of loopback interfaces.255. router network services that can be bound to the loopback interface should be.255 Central(config-if)# end Central# In general. simply assign it an IP address. In addition to physical interfaces. we say that the service is bound to that interface. (When a service is configured to use the loopback interface as its source.) To create a loopback interface. called loopback interfaces. Central(config)# interface loopback0 Central(config-if)# description Main loopback interface Central(config-if)# ip address 14. because the address of the loopback interface is fixed. hiding administrative aspects of your network from potential attackers is usually good practice. Routers and other network devices communicate using various management protocols. such as routing protocols. Also. It means that IP packets generated by the router will have the loopback interface’s address as their source address. Cisco IOS routers have the ability to define internal virtual interfaces. mainly because the source address for some services can vary.

The example below shows how to disable login on the auxiliary port (login to enable mode first): Central# config t Enter configuration commands. To set the router's message-of-the-day banner use the command banner motd delimiter message delimiter. The delimiter can be any single character. The connection to the console port should not be left logged in. See also [7] for more discussion of banners and their content. Each authorized user should log in using their own account (for more details. the router will log him or her out automatically. see the Accounts sub-section below). otherwise you will be locked out of the console. The syntax for creating a local user is username name privilege level password string.1c 59 . Central(config)# username brian privilege 1 password g00d+pa55w0rd Central(config)# line con 0 Central(config-line)# login local Central(config-line)# end Central# The auxiliary port. Configure the console line to time out EXEC sessions. one per line. Router Central.Advanced Security Services the device being subject to monitoring. Central# config t Enter configuration commands. and perhaps a statement threatening prosecution. If you do not already have users accounts set up. should be disabled. Version 1. The example below shows how to set up the console line to enforce a five-minute timeout. Consult your organization’s legal staff or general counsel for suitable language to use in your legal notice. then create at least one before setting the console to use local login. Central(config)# line con 0 Central(config-line)# transport input none Central(config-line)# exec-timeout 5 0 Central(config-line)# exit Central(config)# End with CNTL/Z. if at all possible. but that terminal (or computer) should be standalone. A proper legal notice protects the ability of the owning organization to pursue legal remedies against an attacker.0 and earlier only). so that if an administrator forgets to log out. It is okay to leave a connection to the console port attached all the time. and protected from unauthorized access. The example below shows how to create an account with a password and set console login. The console (con) port is the default location for performing router management and configuration. End with CNTL/Z. one per line. Router model and location information should never be included. in the sample network diagram (Figure 4-1). has no need for the aux port. Note that you must create at least one user account. Be careful not to provide any information in the banner message that should not be shared with the general public. the command transport input none prevents remote access to the console port via reverse-telnet (on IOS 12. Apply the command login local to the console line to enforce user log. Do not include any network architecture or device information in the banner message.

and Sections 5.1.2 and 5.6 for details on configuring remote administration.1. one per line. Most versions of IOS have five virtual terminals. do not perform them while logged into the router via Telnet. 64.] South# config t Enter configuration commands. these connections are called virtual terminal lines. or even more. South(config)# no access-list 90 South(config)# access-list 90 deny any log South(config)# line vty 0 4 South(config-line)# access-class 90 in South(config-line)# transport input none South(config-line)# login local South(config-line)# exec-timeout 0 1 South(config-line)# no exec South(config-line)# end South# End with CNTL/Z.1c . It is important to know how many virtual terminals your IOS version has.3 for cryptographic mechanisms for protecting the remote administration connections. Login on the virtual terminal lines should be disabled if remote administration is not absolutely necessary. If the auxiliary port is required for a second local serial connection then configure it as shown below. If remote administration is necessary.Router Security Configuration Guide Central(config)# line Central(config-line)# Central(config-line)# Central(config-line)# Central(config-line)# Central(config-line)# aux 0 transport input none login local exec-timeout 0 1 no exec exit Section 4.5 discusses configuration of the auxiliary port if it is required for a modem. and 60 Version 1. create an access list and apply it to the virtual terminal lines. as shown in the example below. Some IOS versions (including the versions designated “Enterprise”) may have 15. Central(config)# line Central(config-line)# Central(config-line)# Central(config-line)# Central(config-line)# Central(config-line)# Central# aux 0 exec-timeout 5 0 login local transport input none exec end VTYs and Remote Administration One primary mechanism for remote administration of Cisco routers is logging in via Telnet or SSH. numbered 0 through 4. Remote administration without encryption is inherently dangerous because anyone with a network sniffer on the right LAN segment can acquire the router passwords and would then be able to take control of the router. [Note: perform these commands only when connected to the aux or console port. or use the command transport input none. To disable network virtual terminal connections to the router. see Section 4.

) South# config t Enter configuration commands. If the router has more vtys than you need. Cisco IOS comes with 2 predefined user levels.simply delete 5. you cannot delete VTYs 0 through 4. Every IOS command is pre-assigned to either level 1 or level 15. one per line. If you do not know how many vtys your router supports.1c 61 . or delete them with the configuration mode command no line vty. you would configure all of the vtys on the router identically. (Note: on most IOS versions. End with CNTL/Z. South# show line vty 0 ? <1-935> Last Line range summary Quick line status summary | Output modifiers <cr> South# show line vty 0 935 Tty Typ Tx/Rx A Modem Roty AccO AccI 66 VTY 67 VTY 68 VTY 69 VTY 70 VTY 71 VTY 72 VTY South# Uses Noise Overruns Int 0 0 0/0 0 0 0/0 0 0 0/0 0 0 0/0 0 0 0/0 0 0 0/0 0 0 0/0 - The seven lines of output from the ‘show line’ command indicate that the router South has seven virtual terminals.Advanced Security Services to explicitly configure all of them securely. in terms of security. By default Cisco provides EXEC (level 1) with a few commands which may. two more than the default complement of five. you can list them using the command show line vty in the manner shown below. Normally. The transcript below shows how to delete the extra two vtys on the router South .6 for more details). South(config)# no line vty 5 South(config)# exit South# show line vty 0 935 Tty Typ Tx/Rx A Modem Roty AccO AccI Uses Noise Overruns Int 66 VTY 0 0 0/0 67 VTY 0 0 0/0 68 VTY 0 0 0/0 69 VTY 0 0 0/0 70 VTY 0 0 0/0 South# Privileges Cisco IOS provides for 16 different privilege levels ranging from 0 to 15. make more sense being at a higher privilege level. and both 5 and 6 will disappear. The next example Version 1. then disable the extra ones. If the router is configured with aaa newmodel then local or remote AAA can be used for user authorization (see Section 4. User EXEC mode runs at privilege level 1 and “enabled” mode (privileged EXEC mode) runs at level 15.

South# config t Enter configuration commands. Type 5 uses an iterated MD5 hash which is much stronger. Passwords There are two password protection schemes in Cisco IOS. It is also possible to set up intermediate privilege levels. Central(config)# Central(config)# Central(config)# Central(config)# Central(config)# Central(config)# Central(config)# Central(config)# Central(config)# privilege exec level 15 connect privilege exec level 15 telnet privilege exec level 15 rlogin privilege exec level 15 show ip access-lists privilege exec level 15 show access-lists privilege exec level 15 show logging ! if SSH is supported. use the enable secret command to set a level password instead (see next sub-section). This could be done by assigning a password to an intermediate level. only use the enable secret command. For example. be very careful about moving too much access down from level 15. First. which in most configurations should be protected better. it will not be used and may give away a system password.. privilege exec level 15 ssh privilege exec level 1 show ip The last line is required to move the show command back down to level 1. Third. and line password commands. once a user has write access they could leverage this to acquire greater access. Cisco recommends using Type 5 encryption instead of Type 7 where possible (see “Configuring Passwords and Privileges” in the IOS 12 Security Configuration Guide). and then assigning particular commands to that privilege level. be very careful about moving any part of the configure command down. Type 7 uses the Ciscodefined encryption algorithm which is known to the commercial security community to be weak. if an attempt was made to do something like this there are a few things to be very careful about. one per line.1c . End with CNTL/Z. do not use the username command to set up accounts above level 1. • To protect the privileged EXEC level as much as possible. like 5 or 10. Type 7 encryption is used by the enable password. do not use the enable password command. But. Even if the enable secret is set do not set the enable password.Router Security Configuration Guide shows how to move the commands to the privileged mode. Deciding which commands to assign to an intermediate privilege level is beyond the scope of this document. an organization might want to set up more than the two levels of administrative access on their routers. username. South(config)# enable secret 2-mAny-rOUtEs South(config)# no enable password South(config)# end South# 62 Version 1. Second. this could cause unexpected security holes in the system.

• RADIUS keys (in 12. Version 1.5. • Make all passwords at least eight characters long. no user account should be created above privilege level 1.3). uppercase letters. Do not set any user or line password to the same value as any enable secret password. Good security practice dictates some other rules for passwords. dates. But user accounts should be created for auditing purposes (see Accounts. digits. In addition to the above password access mechanisms. • Avoid more than 4 digits or same-case letters in a row. and special characters.1 and earlier) • TACACS+ keys (in 12.1c 63 . authorize. and audit users (see Section 4. • If the login command is used to protect a line then the line password command is the only way to set a password on a line. For access and logging reasons the login local method should be used. • Peer router authentication keys (in 12. But if the login local command is used to protect a line then the specified user name/password pair is used. The username command should be used to create individual user accounts at the EXEC level and then the higher privilege levels should be protected with enable secret passwords.1 and earlier) – for more information about routing protocol authentication see Section 4. Then users with a need to work at higher levels would be given the higher privilege level password.Advanced Security Services • Because it is not possible to use Type 5 encryption on the default EXEC login or the username command (prior to IOS 12.4. AAA mechanisms may be used to authenticate. • Enable service password-encryption.6 for details). • Always include at least one of each of the following: lowercase letters. below). Never set any of these secret values to the same string as any other password.5. Some of the more important rules are provided in the following list.1 and earlier) • NTP authentication keys – for more information about NTP security. proper names. • Avoid dictionary words. • The privileged EXEC secret password should not match any other user password or any other enable secret password. addresses. phone numbers.3. • SNMP community strings – for more information about SNMP security see Section 4. this will keep passersby from reading your passwords when they are displayed on your screen. see Section 4. • Be aware that there are some secret values that service passwordencryption does not protect.

5. Remote Internal only with AAA – administration can be performed on the router from a trusted internal network only. Central(config)# service password-encryption Central(config)# username rsmith password 3d-zirc0nia Central(config)# username rsmith privilege 1 Central(config)# username bjones password 2B-or-3B Central(config)# username bjones privilege 1 Central(config)# no username brian Central(config)# end Central# Only allow accounts that are required on the router and minimize the number of users with access to configuration mode on the router.1. Accounts First.6. and remove the local user named ‘brian’. the log message that is generated will include the name of the login account which was used. do not create any user accounts without passwords! When an administrator no longer needs access to the router. remove their account. 3. for a preferred user account mechanism. Central# config t Enter configuration commands. and AAA is used for access control. The example below shows how to create local user accounts for users named ‘rsmith’ and ‘bjones’. Remote External with AAA – administration can be performed with both internal and external connections and uses AAA for access control. Note: enable secret and username passwords may be up to 25 characters long including spaces. Remote Internal only – administration can be performed on the router from the internal network only. 1. See Section 4. which describes AAA. In addition. No Remote – administration is performed on the console only. 64 Version 1. 2. Remote External – administration can be performed with both internal and external connections.Router Security Configuration Guide See [4] for more detailed guidance on selecting good passwords.1c . End with CNTL/Z. give each administrator their own login user name for the router. Remote Access This document will discuss five connection schemes which can be used for router administration. 4. one per line. The login accounts created with the username command should be assigned privilege level 1 (see Passwords. above). 4. When an administrator logs in with a user name and changes the configuration.6.

The five regimes listed above are listed in the order that best protects the router and allows for accounting of router activities. When you use remote administration. anyone with a network sniffer and access to the right LAN segment can acquire the router account and password information. For better security. Cisco has added support for the Secure Shell (SSH) protocol to many versions of IOS 12. The Auxiliary Port As discussed in Section 4. Network Access Remote network connections use the VTY lines to connect to the router. To configure the vtys for remote access do the following: bind the telnet service to the loopback interface.6 describes remote access with AAA.5 the aux port should be disabled. the Internet) should not be used. Consult the Cisco IOS Dial Services guide [6] for information about connecting modems and configuring callback. Central(config)# ip telnet source-interface loopback0 Central(config)# access-list 99 permit 14. Setting up IPSec for remote administration is covered in Section 5. create and apply an access list explicitly listing the hosts or networks from which remote administration will be permitted. IOS callback features should be used. Remote access over untrusted networks (e. with or without AAA.0 and later.3T. 12.9.2. remote administration is inherently dangerous. Section 4.1 log Central(config)# access-list 99 permit 14.2. This section will discuss remote internal only access without AAA. so it is necessary to apply access controls to the aux port.1.6.3 describes how to use SSH for secure remote administration. such as IPSec or SSH. Attackers using simple war-dialing software will eventually find the modem.6 log Central(config)# access-list 99 deny any log Central(config)# line vty 0 4 Central(config-line)# access-class 99 in Central(config-line)# exec-timeout 5 0 Central(config-line)# transport input telnet Central(config-line)# login local Central(config-line)# exec Version 1. and SSH should always be used instead of Telnet whenever possible. This is why remote administration security issues center around protecting the paths which the session will use to access the router. A detailed discussion on setting up modems is beyond the scope of this document.6).1c 65 . all connections to the router should require authentication (using individual user accounts) for access. Only if absolutely required should a modem be connected to the aux port as a backup or remote access method to the router.5.Advanced Security Services As discussed in Section 4. This can be accomplished by using login local (see next sub-section for example) or AAA (see Section 4. As discussed earlier.2.1. and set an exec session timeout. because the user’s password will travel the network in clear text form. Section 5. The security of remote administration can be enhanced by using a protocol that provides confidentiality and integrity assurances. and nearly all IOS releases in 12. unless the traffic is adequately protected.4 and later.g.

Offline editing provides the ability to add comments. if you are going to permit remote administration via Telnet. 4.1c . Central(config)# service tcp-keepalives-in Central(config)# service tcp-keepalives-out Central(config)# exit Central# 4.6 contains details on configuring AAA in an example network. Logistics for Configuration Loading and Maintenance There are two basic approaches for configuration loading and maintenance: online editing and offline editing. Online editing provides for syntax checking but provides limited editing capability and no comments. Using this service does not remove the need for setting an exec-timeout time as recommended above. they can establish further telnet sessions from the router to other devices. it should be disabled as shown below. enable TCP keepalive services. Central(config)# line vty 0 4 Central(config-line)# transport output none Central(config-line)# exit Lastly. Additionally. Authentication is the mechanism for identifying users before allowing access to a network component.1. These services will cause the router to generate periodic TCP keepalive messages.Router Security Configuration Guide Central(config-line)# end Central# The IP access list 99 limits which hosts may connect to the router through the vty ports. privileges. The login local command requires a username and password be used for access to the router (this command will be different if you are using AAA with an authentication server). like rlogin and web.8. thus allowing it to detect and drop orphaned (broken) TCP connections to/from remote systems. Authorization is the method used to describe what a user has the right to do once he has authenticated to the router. Finally.3. They each have advantages and disadvantages. and logging of user activities on a router. and Accounting (AAA) This is Cisco’s new access control facility for controlling access.1. the IP addresses which are allowed to connect must be on an internal or trusted network. Section 4. Accounting is the component that allows for logging and tracking of user and traffic activities on the router which can be used later for resource tracking or trouble shooting. For more details on access lists see Section 4. This is important because the other supported protocols. Cisco IOS supports outgoing telnet as well as incoming. Unless this capability is important for managing your network. Authentication.7. allows for the use 66 Version 1. Authorization. once an administrator or attacker has gained telnet access via a VTY. are less secure and should be avoided. the transport input telnet command restricts the management interface to telnet only.

the system where the configuration files are stored should use the local operating system’s security mechanisms for restricting access to the files. Old and alternative configurations should be stored offline. Cisco configuration save utilities will also not save default values. the show run command will only show those configuration settings which are different from the IOS defaults. But the offline method will leave passwords in the clear.Advanced Security Services of better editors. always use the startup configuration to avoid problems. use configuration management to track changes to your configurations. it is best to type the passwords while on-line (using the console) and then copy the encrypted strings to the offline configuration. then displaying the running configuration. Third.1. When managing configuration files offline there are several security issues. rcp.lgcsUnsgtLaWgskteq. When saving and loading configurations. using the terminal is the best choice for loading a new configuration. ! set the enable secret password using MD5 encryption enable secret 5 $1$fIFcs$D.3 recommends performing local administration. because the saving will not include default values and after an IOS upgrade you may encounter unexpected configuration problems. In this situation it is only necessary to manage the startup configuration since the running configuration is identical. so that if there is a power failure or some other problem the router will restart with the correct configuration. The recommended approach is a hybrid of the two. TFTP. and SCP (available in many releases 12. Also. This is especially true for the enable secret password. with the configuration files offline the files must be transferred to the router in the relatively secure method. maintain the configuration offline by writing it offline (see above). and then copying and pasting the encrypted string into your offline configuration file. FTP (available for IOS 12. It is also important to keep the running configuration and the startup configuration synchronized. if you set passwords in an offline configuration file.1c 67 . The primary mechanisms available are the console terminal.0 and newer). Only authorized router administrators should be given access to the files. tracking the configuration can become very difficult. Only a few lines should be copied at a time so that you can determine that the entire Version 1.8 Local and Remote Administration Section 4. Because each Cisco IOS release changes the default values for some of the commands. Only save off the running configuration for an emergency. Instead.1 later that support SSH). then they will be stored in the clear and transferred in the clear. described below. First. The possible methods for transferring files to a router have increased with newer IOS releases. You can obtain the encrypted string by setting the password manually on the router console. In this case. The example below shows how an encrypted enable secret setting would appear in an off-line configuration file. The configuration files would be stored on the computer attached to the console and the local machine’s copy/paste buffer can be used for transferring the configuration to the router. Second. but provides no syntax checking. and guarantees all settings will be visible. With the online editing.

if you have one available. one per line.364 secs Central# 68 Version 1.352 secs (459 bytes/sec) Central# The next example demonstrates how to load a new configuration file into the startup configuration.Router Security Configuration Guide configuration file is transferred successfully. Central(config)# ip ftp username nsmith Central(config)# ip ftp password 1pace-4ward Central(config)# ip ftp source-interface loopback0 Central(config)# exit Central# copy /erase ftp: startup-config Address or name of remote host []? 14.0 then use the FTP protocol to transfer the configuration files to and from the router.0 then using the console connection or a telnet connection is the best choice for administration. Central# config t Enter configuration commands. performs copy/paste very slowly. one per line. On Linux. On Solaris. End with CNTL/Z.2.9.5516/1024 bytes] 5516 bytes copied in 4.. Set the source interface for FTP to the loopback interface if you have defined one.1 Source filename []? /rtr-backup/central-config Destination filename [startup-config]? Accessing ftp://14. If remote administration is allowed and the IOS is newer then version 12.2.1/rtr-backup/central-config. [Note: the default Windows NT 4.1c . the tip command can be used. The following example shows how to save the startup configuration to a file.1 Destination filename [startup-config]? /rtr-backup/central-config Writing central-config !! 5516 bytes copied in 12.0 serial communication program. End with CNTL/Z. use a better communication program. On Windows NT and 2000.9. such as TeraTerm Pro.9. Erasing the nvram filesystem will remove all files! Continue? [confirm] y [OK] Erase of nvram: complete Loading /rtr-backup/central-config ! [OK . otherwise use the interface closest to the FTP server..2. Hyperterminal. the minicom program is suitable for Cisco local console access. Central# copy running-config startup-config Central# config t Enter configuration commands. The file would again be transferred using the host systems copy/paste buffer to move the text from a file editor to the terminal emulator.] If remote administration is being allowed and the router is running an IOS older than version 12. Central(config)# ip ftp username nsmith Central(config)# ip ftp password 1pace-4ward Central(config)# ip ftp source-interface loopback0 Central(config)# exit Central# copy startup-config ftp: Address or name of remote host []? 14.

Cisco Press. 1999..htm This federal standard includes some good guidelines on choosing passwords that are difficult to guess. Cisco Press. [8] Stewart. O’Reilly & Associates.nist. This is the reference manual and guide for basic IOS configuration tasks. Cisco IOS 12. because it provides integrity and confidentiality protection. B. 1999. Cisco Press. 1999. are less secure than FTP and should not be used for loading or saving router configurations. A very specific guide to configuring many IOS features securely. J. Cisco ISP Essentials. It includes information about configuring logins. 1st Edition.5 for details on using TFTP if required. National Institute of Standards and Technology. This is the reference manual and guide for major security features in IOS 12. modem. especially for initial set-up of a new router. 2000. SANS Institute. Relevant sections include: IOS User Interfaces and File Management. and more. vtys. such as rcp and TFTP. Cisco Press. McGraw-Hill. A. and an in-depth treatment of Unicast RPF. Securing Cisco Routers: Step-by-Step. Configuring Passwords and Privileges. The sections about passwords and warning banners contain very good information. T.5. References [1] Cisco IOS Release 12. A pragmatic and detailed guide to securing Cisco routers. Version 1.Advanced Security Services The other protocols. [5] Greene.9. SCP should be used whenver possible. Hardening Cisco Routers. all with fully worked-out examples.1c 69 . 4. This is the reference manual and guide for serial line.1.0 Configuration Fundamentals. and Traffic Filtering and Firewalls. [7] Akin. J. This detailed Cisco guide for Internet Service Providers includes extensive discussion of routing protocols (especially BGP). 2002. An excellent introduction to basic usage and configuration of IOS routers. and Wright. See Section 4.0 Security Configuration Guide. Relevant sections include: Security Overview. P.. and Smith. and dial-in features.0.. ed. J. [4] “Password Usage” Federal Information Processing Standard Publication 112. 2002. [2] Buckley. [3] Albritton. [6] Cisco IOS Dial Services Configuration Guide.itl. Cisco IOS Essentials. 1985. April available at: http://www.

In many cases. Cisco IOS supports turning a service off entirely.2: enabled 11. Turning off an automatic network feature usually prevents a certain kind of network traffic from being processed by the router or prevents it from traversing the router. As stated in Section 3. chargen. Typical Services. Some of these services are application layer protocols that allow users and host processes to connect to the router. disable it. general security practice for routers should be to support only traffic and protocols the network needs. This is a legacy feature. This is a legacy feature. 3. discard. etc. Turning off a network service on the router itself does not prevent it from supporting a network where that protocol is employed. Required Services. For example. 12.2: enabled Enabled Recommendation CDP is almost never needed.3.1c . Standard TCP network services: echo. and 7.2. Unix user lookup service. etc. allows remote listing of logged in users.0. the router’s bootp server should be disabled. Others are automatic processes and settings intended to support legacy or specialized configurations but which are detrimental to security.3: disabled 11.2. Some of these services can be restricted or disabled. If a particular portion of a network needs a service but the rest does not. Default Enabled 11. or restricting access to particular network segments or sets of hosts. Standard UDP network services: echo. and later versions. most of the services listed below are not needed. This list has been kept short by including only those services and features that are security-relevant and may need to be disabled. Unless it is required for the network to operate. improving security without degrading the operational use of the router. Unauthorized persons don’t need to know this. 4.Router Security Configuration Guide 4. 4. 70 Version 1. Router Network Service Security Cisco routers support a large number of network services at layers 2. a router may support a network where the bootp protocol is employed. disable it. IP source routing should be disabled. disable it explicitly.1. For example.3: disabled 11. then the restriction features should be employed to limit the scope of the service. disable it explicitly. and Security Risks The table below lists some of the services offered on Cisco IOS 11. Table 4-1: Overview of IOS Features to Disable or Restrict Feature Cisco Discovery Protocol (CDP) TCP small servers UDP small servers Finger Description Proprietary layer 2 protocol between Cisco devices. IP source routing is a little-used feature of IP that can be utilized in network attacks. but some other host is acting as the bootp server. In this case.

This is rarely needed and may open a security hole. Feature that allows a packet to specify its own route. remove default community strings and explicitly disable. disable on interfaces to untrusted networks. Router can act as a time server for other devices and hosts. Router will send an ICMP redirect message in response to certain routed IP packets. disable it. disable it. Bootp server Configuration auto-loading PAD service IP source routing Proxy ARP Disabled Enabled Enabled Enabled IP directed broadcast IP unreachable notifications IP mask reply Enabled (11.3 & earlier) Enabled Disabled IP redirects Enabled Maintenance Operations Protocol (MOP) NTP service Enabled (on Ethernet interfaces) Enabled (if NTP is configured) Simple Network Mgmt. Can be helpful in attacks. explicitly disable on interfaces to untrusted networks. otherwise restrict access. otherwise restrict access. Protocol Enabled Domain Name Service Enabled (broadcast) Version 1. Directed broadcast can be used for attacks. Can aid network mapping.25 packet assembler service. Disable if not explicitly needed. Disable this service unless the router is serving as a LAN bridge. Set the DNS server addresses explicitly. disable on interfaces to untrusted networks. or disable DNS lookup. This is rarely used. Can aid IP address mapping.Advanced Security Services Feature HTTP server Description Some Cisco IOS devices offer web-based configuration. Can aid network mapping. Router will send an interface’s IP address mask in response to an ICMP mask request. Default Varies by device Enabled Recommendation If not in use. disable it. part of the DECNet protocol suite. otherwise restrict access. Disable if not explicitly needed. explicitly disable. Routers can support SNMP remote query and configuration. If not in use. explicitly disable. disable it if it is not in use. Router will attempt to load its configuration via TFTP. Legacy management protocol . Packets can identify a target LAN for broadcasts. Service to allow other routers to boot from this one.1c 71 . Router will explicitly notify senders of incorrect IP addresses. Router will support X. Router will act as a proxy for layer 2 address resolution. Routers can perform DNS name resolution. If not in use.

one per line.2.. Open Monday. To turn off CDP entirely. April 3.9. Central# ! if connect succeeds.1c .250 daytime Trying 14. use the commands shown below in global configuration mode..2.250 closed by foreign host] Central# config t Enter configuration commands.9.3 and 12.9. one per line.250 daytime Trying 14..9. 2000 11:48:39-EDT [Connection to 14. The example below shows how to test whether the TCP small servers are running. 13 . and then disable it on each interface where it is not needed using the no cdp enable command in interface configuration mode. To enable CDP use the cdp run command in global configuration mode.2. it is not necessary for routers to support these services. In virtually all cases.2. Central(config)# no cdp run Central(config)# exit Central# show cdp % CDP is not enabled Central# End with CNTL/Z.Router Security Configuration Guide 4. and how to disable the TCP and UDP small servers.250. it can be enabled and disabled for each interface.2. % Connection refused by remote host Central# 72 Version 1. and they should be disabled. TCP and UDP Small Servers The TCP and UDP protocol standards include a recommended list of simple services that hosts should provide. CDP The Cisco Discovery Protocol is a proprietary protocol that Cisco devices use to identify each other on a LAN segment.9. and is considered deleterious to security. In the unlikely event that CDP is needed for part of a network. It is useful only in specialized situations. How to Disable Unneeded Features and Services Each sub-section below describes how to disable or restrict particular services and features under Cisco IOS 11. then tcp-small-servers are enabled Central# connect 14.2.250..2. 13 . Central(config)# no service tcp-small-servers Central(config)# no service udp-small-servers Central(config)# exit Central# connect 14. Central# config t Enter configuration commands. End with CNTL/Z.

79 .250. then its use should be restricted as follows.2. Therefore. Central(config)# no ip finger Central(config)# no service finger Central(config)# exit Central# connect 14. Central# config t Enter configuration commands.250.. If web-based administration is examined and found necessary for network operations.2.2. % Connection refused by remote host Central# End with CNTL/Z.Advanced Security Services Finger Server The IOS finger server supports the Unix ‘finger’ protocol. then it should be disabled as shown below. % Connection refused by remote host Central# HTTP Server Newer Cisco IOS releases support web-based remote administration using the HTTP protocol. Version 1.2. Central# connect 14. Typically. If web-based remote administration is not needed. configuring. web-based administration imposes the requirement that users log in at full (level 15) privilege.2.. they are a viable mechanism for monitoring.2..1c 73 .9.250 www Trying 14.250 finger Trying 14.9.250 finger Trying 14.6 00:00:00 goldfish *131 vty 1 idle 00:00:00 central [Connection to 14. Further. one per line.9.250. it is important to note that both Telnet and web-based remote administration reveal critical passwords in the clear.2.9. The example below shows how to test and disable the finger server.9. the show users command may be used to list the logged in users.. On a Cisco router.. web-based remote administration should be avoided.9. access restricted. 79 . 80 . Open This is the CENTRAL router. Line User Host(s) Idle Location 130 vty 0 14. End with CNTL/Z. which is used for querying a host about its logged in users.250 closed by foreign host] Central# config t Enter configuration commands.9. one per line. users who are not authorized to log in to the router have no need to know who is logged in. While the web access features are fairly rudimentary on most Cisco router IOS releases.9. Central(config)# no ip http server Central(config)# exit Central# connect 14. and attacking a router.. Web-based remote administration is useful primarily when intervening routers or firewalls prevent use of Telnet for that purpose.2. However.

6. Central(config)# ! Add web admin users. End with CNTL/Z. as described in Section 5. The router’s web server will use HTTP basic authentication to demand a username and password (unfortunately. but IPSec is preferred.5. (Some recent versions of IOS can support SSL (HTTPS) for protecting administrative HTTP traffic. • Create and apply an IP access list to limit access to the web server.2.18 log Central(config)# access-list 29 permit 14. use the commands shown below. as discussed in Section 4.6.0 network and the host 14.1c . use AAA user access control as described in Section 4. Central# config t Enter configuration commands.255 log Central(config)# access-list 29 deny any log Central(config)# ! Apply the access list then start the server Central(config)# ip http access-class 29 Central(config)# ip http server Central(config)# exit Central# If possible.2. AAA will give more control and better audit.2. The example below illustrates each of these points. • Configure and enable syslog logging as described in Section 4.18 only. and offers an attacker the ability to download a copy of a router’s IOS software.) Bootp Server Bootp is a datagram protocol that is used by some hosts to load their operating system over the network.0 0. primarily for other Cisco hardware. HTTPS is better than plain HTTP.6. This guide does not provide instructions on configuring SSL. This facility is intended to support a deployment strategy where one Cisco router acts as the central repository of IOS software for a collection of such routers. Cisco IOS does not yet support the superior HTTP digest authentication standard). To disable bootp service. Administrators will be allowed to connect from the 14.2. consult the Cisco IOS documentation. If possible. protect the HTTP traffic by setting up IPSec. 74 Version 1. Cisco routers are capable of acting as bootp servers. In practice. one per line. Central(config)# no ip bootp server Central(config)# exit End with CNTL/Z. bootp is very rarely used. Access lists are described in Section 4.3.1.Router Security Configuration Guide • Set up usernames and passwords for all administrators. then turn on http auth Central(config)# username nzWeb priv 15 password 0 C5-A1rCarg0 Central(config)# ip http auth local Central(config)# ! Create an IP access list for web access Central(config)# no access-list 29 Central(config)# access-list 29 permit host 14.2. one per line. Central# config t Enter configuration commands.

The example below shows how to disable proxy ARP on four Ethernet interfaces. Unless a network depends on source routing.25.g. even on interfaces that are currently idle.1c 75 . Disable it on each interface where it is not needed. Because it breaks the LAN security perimeter. Cisco routers perform proxy ARP by default on all IP interfaces. a standalone lab network). Disable it from global configuration mode as shown below. and should be considered only on a network that is wholly trusted (e. PAD Service The packet assembler/disassembler (PAD) service supports X. responding to ARP queries on selected interfaces and thus enabling transparent access between multiple LAN segments. Central(config)# no ip source-route Central(config)# Proxy ARP Network hosts use the Address Resolution Protocol (ARP) to translate network addresses into media addresses.Advanced Security Services Configuration Auto-Loading Cisco routers are capable of loading their startup configuration from local memory or from the network. proxy ARP should be used only between two LAN segments at the same trust level. using the interface configuration command no ip proxy-arp . effectively extending a LAN at layer 2 across multiple segments. This service is on by default. Cisco routers normally accept and process source routes. This feature is used in several kinds of attacks. A Cisco router can act as intermediary for ARP. but it is not needed unless your router is using X. Normally. one per line. ARP transactions are confined to a particular LAN segment. Loading from the network is not secure. The example below shows how to disable IP source routing.25 links. and only when absolutely necessary to support legacy network architectures. Version 1. Central(config)# no boot network Central(config)# no service config Central(config)# exit Central# End with CNTL/Z. Central(config)# no service pad Central(config)# IP Source Routing Source routing is a feature of IP whereby individual packets can specify routes. Central# config t Enter configuration commands. Explicitly disable loading the startup configuration from the network using the commands shown below. it should be disabled on all the net’s routers. This service is called proxy ARP.

Mask Replies The Internet Control Message Protocol (ICMP) supports IP traffic by relaying information about paths. and network conditions. Automatic generation of these messages should be disabled on all interfaces. routes.2. Central# config t Enter configuration commands. Cisco routers automatically send ICMP messages under a wide variety of conditions. Central(config)# interface eth 0/0 Central(config-if)# no ip proxy-arp Central(config-if)# exit Central(config)# interface eth 0/1 Central(config-if)# no ip proxy-arp Central(config-if)# exit Central(config)# interface eth 0/2 Central(config-if)# no ip proxy-arp Central(config-if)# exit Central(config)# interface eth 0/3 Central(config-if)# no ip proxy-arp Central(config-if)# end Central# IP Directed Broadcast Directed broadcasts permit a host on one LAN segment to initiate a physical broadcast on a different LAN segment. one per line. IP Unreachables. Explicitly disable directed broadcasts on each interface using the interface configuration command no ip directed-broadcast as shown in the example in the next subsection. especially interfaces that are connected to untrusted networks.15. and ‘Mask Reply’. Central(config)# interface eth 0/0 Central(config-if)# no ip unreachables Central(config-if)# no ip redirects Central(config-if)# no ip mask-reply Central(config-if)# no ip directed-broadcast Central(config-if)# end Central# 76 Version 1. End with CNTL/Z.1.250 YES NVRAM up up Ethernet0/1 14. ‘Redirect’.250 YES NVRAM up up Ethernet0/2 unassigned YES unset down down Ethernet0/3 unassigned YES unset down down Central# config t Enter configuration commands. The example below shows how to turn them off for an interface. This technique was used in some old denialof-service attacks. Redirects.Router Security Configuration Guide Central# show ip interface brief Interface IP-Address OK? Method Status Protocol Ethernet0/0 14. one per line.1c .9. and the default Cisco IOS configuration is to reject directed broadcasts. End with CNTL/Z. Three ICMP messages are commonly used by attackers for network mapping and diagnosis: ‘Host unreachable’.

• Disable SNMP system processing. The example below shows how to disable SNMP by implementing these recommendations. North# show ip interface brief Interface IP-Address OK? Method Status Protocol Ethernet0/0 14.Advanced Security Services MOP The Maintenance Operations Protocol (MOP) was used for system utility services in the DECnet protocol suite. use an access list. then all routers on that network should be configured to securely participate in it. There are several different versions of SNMP.250 YES NVRAM up up North# config t Enter configuration commands.20 YES NVRAM up up Ethernet1/0 14. It is enabled by default on Ethernet interfaces in some versions of IOS.1. one per line.2. Disable it on each Ethernet interface as shown below. End with CNTL/Z. It starts with listing the current configuration to find the SNMP Version 1. SNMP Services The Simple Network Management Protocol (SNMP) is the standard Internet protocol for automated remote monitoring and administration.5. If a network has a deployed SNMP infrastructure in place for administration. Central(config)# interface eth 0/0 Central(config-if)# no mop enabled Central(config-if)# end NTP Service Cisco routers and other hosts use the Network Time Protocol (NTP) to keep their time-of-day clocks accurate and in synchrony.10. North(config)# interface eth 0/0 North(config-if)# ntp disable North(config-if)# exit North(config)# interface eth 1/0 North(config-if)# ntp disable North(config-if)# end North# Disabling NTP on an interface will not prevent NTP messages from traversing the router. If NTP services are not available on the network. If possible.1. as described in Section 4. all SNMP facilities on all routers should be disabled using these steps: • Explicitly unset (erase) all existing community strings.1c 77 . then disable NTP as shown below. In the absence of a deployed SNMP scheme. with different security properties. as discussed in Section 4. configure all routers as part of an NTP hierarchy. To reject all NTP messages at a particular interface. • Disable SNMP system shutdown and trap features.3.

using the 78 Version 1. Central(config)# ! erase old community strings Central(config)# no snmp-server community public RO Central(config)# no snmp-server community admin RW Central(config)# Central(config)# ! disable SNMP trap and system-shutdown features Central(config)# no snmp-server enable traps Central(config)# no snmp-server system-shutdown Central(config)# no snmp-server trap-auth Central(config)# Central(config)# ! disable the SNMP service Central(config)# no snmp-server Central(config)# end The last command in the example.Router Security Configuration Guide community strings. snmp-server community public RO snmp-server community admin RW Central# Central# config t Enter configuration commands. one per line.255. see Section 4. For information on setting up and using SNMP securely. is to list the established SNMP community strings and explictly unset them as shown above.255.3. turn off DNS name resolution as shown below.. one per line. By default. so you may want to use IOS output filtering to display only the lines related to SNMP (under IOS 12.255.1c . Router Name and DNS Name Resolution Cisco IOS supports looking up host names with DNS. North(config)# no ip domain-lookup North(config)# end End with CNTL/Z. then explicitly set the name server addresses using the global configuration command ip name-server addresses. but they can still be there! The safest way to ensure that SNMP is really unavailable to an attacker. shuts down all SNMP processing on the router. It is also a very good idea to give the router a name. DNS name resolution should be enabled on a router only if one or more trustworthy DNS servers are available. you must simply list the entire configuration and inspect it visually). When SNMP processing is shut down.. The configuration listing is often quite long. End with CNTL/Z. and will remain so.0 and earlier. some SNMP configuration statements will not appear in any listing of the running configuration. If one or more name servers are available on the network.5. Central# show running-config | include snmp Building configuration. note that SNMP must be enabled in order for the SNMP community strings to appear in the configuration listing. and you want to be able to use names in IOS commands. no snmp-server. IOS sends DNS name queries to the broadcast address 255. In general. North# config t Enter configuration commands. If you do not want your router to send queries.

This helps discourage unauthorized use of extra interfaces.1.1. one per line. router# config t Enter configuration commands. End with CNTL/Z. The example below shows how to set the router name. as described above. see Section 4. use the command shutdown in interface configuration Central(config)# end Central# 4. 4. For more information about NTP and SNMP security configuration.testnet. Disable Unused Interfaces It is a good idea to explicitly shut down (disable) unused interfaces on your router.4. Configuration Example The configuration listing below shows the configuration commands for disabling typical unneeded services. Central# config t Enter configuration commands. the name you give to the router will appear in the prompt. and set up a main and backup DNS server address for the router Central. and enforces the need for router administration privileges when adding new network connections to a router. ! ----.2. This sample is formatted as it would appear in a configuration text file stored on a host for download to the router Central. To disable an interface. use the config command ip domain-name domain as shown below.Advanced Security Services command hostname. To set a domain name. Central(config)# interface eth0/3 Central(config-if)# shutdown Central(config-if)# end Central# End with CNTL/ Central(config)# ip domain-name testnet.3).2 14.9.2. Setting a domain name is also necessary for using SSH (see Section 5. one per line.2.IP and network services section no cdp run no ip source-route no service tcp-small-serv no service udp-small-serv no ip finger no service finger no ip bootp server Version 1. which will be used as part of the fullyqualified host name of the router and any unqualified name lookups. Central(config)# ! full name of this router: Central. router(config)# hostname Central Central(config)# ip name-server 14.3.1 Central(config)# ip domain-lookup Central(config)# You can also set a default DNS domain name.1c 79 .

0.1c .SNMP Section (for totally disabling SNMP) ! disable SNMP trap and system-shutdown features no snmp-server enable traps no snmp-server system-shutdown no snmp-server trap-auth ! turn off SNMP altogether no snmp-server ! ----.2.9.Per-interface services section interface eth 0/0 description Outside interface to 14.1.Router Security Configuration Guide no no no no service pad ip http server ip name-server ip domain-lookup ! ----.Boot control section no boot network no service config ! ----.0/24 net no ip proxy-arp no ip directed-broadcast no ip unreachable no ip redirect no mop enabled ntp disable interface eth 0/2 description Unused interface no ip proxy-arp no ip directed-broadcast no ip unreachable no ip redirect no mop enabled ntp disable shutdown interface eth 0/3 description Unused interface no ip proxy-arp no ip directed-broadcast no ip unreachable no ip redirect no mop enabled ntp disable shutdown interface loopback0 80 Version 1.0/16 net no ip proxy-arp no ip directed-broadcast no ip unreachable no ip redirect no mop enabled ntp disable interface eth 0/1 description Inside interface to 14.

Cisco Security Architectures.2. Cisco IOS 12. Version 1. September 1999. 1999. Cisco Press. Short article with some good advice on features to turn off. Cisco Internetworking Case Studies. 1999. available under: http://www. plus excellent coverage of access lists. [6] Held. [4] Buckley. New York. Cisco Systems.1c 81 .3. B. [2] “Cisco IOS Version 12. 1999. though.php?p=55&a=10 A concise and readable article with practical advice on setting up a router at a boundary between a trusted and untrusted network.” Phrack Magazine.” National Y2K Information Coordination Center. G. 1998. K. includes common-sense measures to take on routers running IOS 11. September 1999.phrack. The section on “IP Addressing and Services” includes information about several of the IP services described in this section. Good overview of Cisco router and TCP/IP [3] “Increasing Security on IP Networks.” .0 Configuration Fundamentals.0 Security Configuration. Cisco Press. The sections on “Performing Basic System Management” and “Monitoring the Router and Network” include valuable advice on how to configure basic features and Very helpful article from Cisco. McGraw-Hill. References [1] Eldridge. Part Available from Cisco’s web site.5. 9 Issue 55.Advanced Security Services description Loopback interface for service bindings no ip proxy-arp no ip directed-broadcast no ip unreachable no ip redirect ntp disable 4. A. “Building Bastion Routers Using Cisco IOS. available at: http://www. Can’t seem to find it on the web anymore. [5] Cisco IOS Network Protocols Configuration Guide. and Hundley. Vol.

al. ed. “HTTP Authentication: Basic and Digest Access Authentication”. June 1999. 82 Version 1. including several of the ones discussed in this section.1c . J. “Requirements for IP Version 4 Routers”.. RFC 2617.Router Security Configuration Guide [7] Franks. RFC 1812. This comprehensive standard describes the services that routers must or may provide. [8] Baker. F. June 1995. et. The standard for HTTP basic authentication used for access control by Cisco IOS web remote administration.

and for filtering traffic passing through the router. Version 1. An access list consists of one or more rules. and for other packet identification tasks. Syntax The basic structure for an access list rule is shown below. Access Control Lists. source or destination TCP/UDP ports.3. source or destination IP addresses. Secure configuration of Cisco routers makes very heavy use of access lists. Concepts Access lists on Cisco routers provide packet selection and filtering capabilities. Extended access lists can permit or deny packets based on their protocols. IPSec. 4. Only standard IP access lists can be applied to SNMP. and what kind of access list it is. specifies which packets match the rule. with some extensive examples. there are two types of access lists available: standard and extended. Both standard and extended IP access lists can be applied to router interfaces. Extended access lists also support selective logging. and Rate Limiting Cisco IOS uses access lists to separate data traffic into that which it will process (permitted packets) and that which it will not process (denied packets). access-list list-number {deny | permit} condition The access list number tells Cisco IOS which access list the rule should be a part of. vty lines (for remote access).1c 83 .1.Advanced Security Services 4. The condition field. Filtering. Conditions typically involve protocol information and addresses. or ICMP or IGMP message types. but do not involve application-level information. permit permits access if the condition is matched. for restricting access to services on the router itself. which is different for each kind of access list. This section gives a moderately detailed description of access list syntax. For IP traffic. The following is the syntax for a statement (rule) in a standard IP access list: access-list list-number {deny | permit} source [source-wildcard] [log] where list-number is the number of the access list and can be any decimal number from 1 to 99. source-wildcard is the wildcard bits to be applied to the source. Standard access lists only allow source IP address filtering. and many router features. deny denies access if the condition is matched. routing protocols. source is the IP address of the network or host from which the packet is being sent.3.

This feature allows you to refer to an access list by a descriptive name instead of by number. Note that logging for IP standard access lists is supported only in IOS 12. The following is simplified syntax for a statement in an extended IP access list: access-list list-number {deny | permit} protocol source source-wildcard source-qualifiers destination destination-wildcard destination-qualifiers [ log | log-input] where list-number is the number of the access list and can be any decimal number from 100 to 199. tcp or udp. The keyword any can be used in place of destination and destination-wildcard. ospf. The syntax for defining an IP access list by name is shown below. causes a message about the packet that matches the statement to be logged. source-wildcard is the wildcard bits to be applied to the source. The keyword any can be used in place of source and source-wildcard. icmp. ipinip. After the list is defined by name. gre. including port numbers and other protocol-specific information.0 and later. log. It can be one of the following keywords: eigrp.Router Security Configuration Guide The optional keyword log may be applied to log matches to the rule. if present. ip.) source is the IP address of the network or host from which the packet is being sent. igmp.1). deny denies access if the condition is matched. destination is the IP address of the network or host to which the packet is being sent. It also provides a convenient way to build lists on-line. source-qualifiers are optional details on the packet source. Cisco has also created an alternative called named IP access lists for both standard and extended lists. nos. Or it can be an integer in the range 0 to 255 representing an IP protocol number. including port numbers and other protocol-specific information.1c . permit permits access if the condition is matched. (Some protocols allow further qualifiers: source or destination ports can be specified for tcp or udp. and message types can be specified for icmp or igmp. destination-qualifiers are optional details on the packet destination. protocol is the name or number of an IP-related protocol.5. destination-wildcard is the IP address wildcard bits to be applied to the destination. you can add statements beginning with either the permit or deny 84 Version 1. and log-input causes a message that includes the interface (logging is described in Section 4. igrp.

1.1.1. After the permit or deny keyword the syntax is the same as defined above for either the standard list or the extended list. It is possible however to apply two access lists (one for each direction) for router interfaces. and any TCP traffic from the Internet to all hosts in the 14.1.2. some UDP traffic) the access list will only affect approximately half the packets. vty lines and routing protocols.1. they should not be allowed across the router in either direction or to the router. Otherwise. ip access-list {standard | extended} name where standard specifies a standard IP access list.0/16 network. North(config)# North(config)# North(config)# North(config)# access-list access-list access-list access-list 140 140 140 140 permit deny permit deny udp udp tcp ip any any any any host 14.255 North(config-ext-nacl)# deny ip any any log North(config-ext-nacl)# exit General Recommendations Refer to the two tables in Section 3. an access list with no permit statements will block all network traffic wherever it is applied. The diagram below shows how access lists work when applied to router interfaces.2. For any connection that requires two-way interaction (e.0. The first table lists those services that should be completely blocked at the router. The second table lists those services on the internal network or on the router that should not be accessible by external clients.0 0. all TCP traffic.Advanced Security Services keyword.255 any log The example below shows the same list as a named IP access list.255.1.1. The name cannot contain spaces or punctuation and must begin with an alphabetic character.0.0. Note that an access list is applied to packets traveling in one direction only.2 that present common services to restrict because they can be used to gather information about an internal network or they have weaknesses that can be exploited.1.g.2 eq 53 any log 14.0 0. Version 1.0.1c 85 . eq 53 North(config-ext-nacl)# deny udp any any log North(config-ext-nacl)# permit tcp any 14. North(config)# ip access-list extended border-filter-14 North(config-ext-nacl)# permit udp any host 14.. Syntax Examples The example below shows how to create a small extended IP access list that permits DNS traffic to the address 14. In each access list there must be at least one permit statement. extended specifies an extended IP access list. name is the name of the access list. using the router East as an example.

2.6.0 network Trash Trash permit Outbound Access List Routing Fabric Outbound Access List permit Figure 4-2: Conceptual Model for Access Lists on Interfaces Use the log keyword at the end of each deny statement in each extended access list.0.255 any East(config)# access-list 102 deny ip any any log-input Add the following statements at the end of each extended IP access list to deny and to log any packets that are not permitted.0. Logs of denied packets can be useful for detection and analysis of probes and attacks against a network.2. whenever you need to change an access list. you cannot easily modify access lists.250 14. then use the qualifier log-input instead of log.20 East Eth1 14. Thus.0 0. it is best to build it offline on a separate computer.0.1c . If you might apply the same access list to more than one interface.Router Security Configuration Guide 14. which is usually sufficient to identify the provenance of the traffic. due to limited editing capability on the Cisco router. East(config)# access-list 100 deny tcp East(config)# access-list 100 deny udp East(config)# access-list 100 deny ip any any any any any range 0 range 0 range 0 range 0 any log 65535 65535 log 65535 65535 log Finally. These statements include the entire port ranges for TCP and UDP explicitly.1.0/16 Eth0 14.2. Log messages generated by access lists are at log level 6 ‘Informational’. East(config)# access-list 102 permit ip 14. Access list log messages always include the access list number.0.1. This will guarantee that the router will log the values for the source and destination ports for TCP and UDP traffic. Since the original access list is 86 Version 1. When the access list is ready you can cut and paste the access list via a connection to the router.6.6.1. This feature will provide valuable information about what types of packets are being denied.2. as shown in the example below.0/24 Interface Eth0 Interface Eth1 permit 14.0 network Inbound Access List deny permit Inbound Access List deny 14.6.1.

more reliable. Remote Login (Telnet) Service There are a number of methods to filter access to the router itself: vty lines. SNMP servers and routing protocols.1c 87 .18 to connect to the router East via Telnet.18 any eq 23 log East(config)# access-list 105 deny ip any any log East(config)# line vty 0 4 East(config-line)# access-class 105 in East(config-line)# end SNMP Service A Cisco router can be configured to act as a client for SNMP. then apply it to the interface or service where you need it. For more information about services on the router.2. East(config)# access-list 105 permit tcp host 14. Below is an example of how to clear an access list.6.2.2 and 4. SNMP should be used only on internal or protected networks.6 and 14.6.2. The following example shows the configuration of an extended IP access list that is applied to the vty lines. The list denies all other connections. Typically.6.6 East(config)# access-list 75 deny any log East(config)# snmp-server community N3T-manag3m3nt ro 75 For more information about SNMP configuration. When SNMP service is enabled on a router.3. Thus. It also logs all successful and unsuccessful connections. and how to disable unneeded ones. and more efficient to use the specialized facilities that IOS makes available to apply access controls directly to the services themselves.2. your changes will take place instantly. you must purge it before adding the updated access list.2. It is safest to build an access list completely first. Most of the time. The following example shows the configuration of a standard IP access list that is applied to a snmp server. The vty lines are used for remote access to the router. route table. The list denies all other connections. 4. East(config)# no access-list 100 Be careful when clearing and rebuilding an access list that is in use. This access list allows the host with IP address 14.6 to gather SNMP information from the router. Version 1. possibly dropping traffic or even leaving the router vulnerable. it is typically easier.5.2.6. see Sections 4. network management tools can use it to gather information about the router configuration. This simple IP access list allows the hosts with IP addresses Versions 1 and 2 of SNMP are not considered secure due to the lack of strong authentication.6 any eq 23 log East(config)# access-list 105 permit tcp host 14. see Section 4.6. East(config)# access-list 75 permit host 14. While it is possible to incorporate access controls for these services into the access lists placed on interfaces.3. Filtering Traffic to the Router Itself Access lists are used in a variety of ways to control access to services on the router itself.2. a router administrator telnets to one of the vty lines. and more. traffic load.Advanced Security Services still on the router.

Router Security Configuration Guide

Routing Service
Communications between routers for routing table updates involve routing protocols. These updates provide directions to a router on which way traffic should be routed. You can use access lists to restrict what routes the router will accept (in) or advertise (out) via some routing protocols. The distribute-list acl-num out command is used to restrict routes that get distributed in routing updates, while the distribute-list acl-num in command may be used used to filter routes that will be accepted from incoming routing updates. The following example shows the configuration of a standard IP access list applied with the EIGRP routing protocol. With the access list applied, router South will not advertise routes to the network.
South(config)# access-list 10 deny South(config)# access-list 10 permit any South(config)# router eigrp 100 South(config-router)# distribute-list 10 out South(config-router)# end South#

Access lists can be used for general filtering of routing updates with distance-vector routing protocols like RIP, EIGRP, and BGP. With link-state routing protocols like OSPF, access lists can be used only for some specialized kinds of filtering. For more information about this topic, see Section 4.4.

4.3.3. Filtering Traffic through the Router
The following examples illustrate methods to protect the router or the internal network from attacks. Note: these separate examples should not be combined into one access list because the result would contain contradictions. In the next section an example configuration file is presented that shows one way to combine these methods into access lists. Refer to the network diagram in Figure 4-1 to understand the example interfaces, their IP addresses and the corresponding access lists.

IP Address Spoof Protection
The filtering recommendations in this sub-section are applicable to border routers, and most interior routers. With backbone routers, it is not always feasible to define ‘inbound’ or ‘outbound’. Additional guidance and recommandations maybe found in [6] and [11].

Inbound Traffic
Do not allow any inbound IP packet that contains an IP address from the internal network (e.g.,, any local host address (, the link-local DHCP default network (, the documentation/test network (, or any reserved private addresses (refer to RFC 1918) in the source field. Also, if your network does not need multicast traffic, then block the IP multicast address range


Version 1.1c

Advanced Security Services

( Apply this access list to the external interface of the router, as shown in the transcript below.
East(config)# no access-list 100 East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip any East(config)# access-list 100 deny ip host any East(config)# access-list 100 permit ip any East(config)# access-list 100 deny ip any any log East(config)# interface eth0 East(config-if)# description External interface to net East(config-if)# ip address East(config-if)# ip access-group 100 in East(config-if)# exit East(config)# interface eth1 East(config-if)# description Internal interface to net East(config-if)# ip address East(config-if)# end log log log log log log log log log log

Outbound Traffic
Do not allow any outbound IP packet that contains an IP address other than a valid internal one in the source field. Apply this access list to the internal interface of the router. See example rules below.
East(config)# no access-list 102 East(config)# access-list 102 permit ip any East(config)# access-list 102 deny ip any any log East(config)# interface eth 0/1 East(config-if)# description "internal interface" East(config-if)# ip address East(config-if)# ip access-group 102 in

On most Cisco routers, IOS 12 offers another mechanism for IP address spoof protection: IP unicast reverse-path forwarding verification. Though specialized, and not suitable for all networks, this facility offers good performance and ease of maintenance. Section 4.4.7 shows how to set up reverse-path forwarding verification on routers that support it.

Exploits Protection
This sub-section describes how to use access lists to defeat or discourage several common attacks using IOS traffic filtering capabilities.

Version 1.1c


Router Security Configuration Guide

TCP SYN Attack
The TCP SYN Attack involves transmitting a volume of connections that cannot be completed at the destination. This attack causes the connection queues to fill up, thereby denying service to legitimate TCP users. The following discussion shows two different approaches. External Access Blocked The access list rules shown below will block packets from an external network that have only the SYN flag set. Thus, it allows traffic from TCP connections that were established from the internal network, and it denies anyone coming from any external network from starting any TCP connection.
East(config)# access-list 106 permit tcp any established East(config)# access-list 106 deny ip any any log East(config)# interface eth 0/0 East(config-if)# description External interface East(config-if)# ip access-group 106 in

Limiting External Access with TCP Intercept The access list rules shown below will block packets from unreachable hosts using the TCP intercept feature; thus, it only allows reachable external hosts to initiate connections to a host on the internal network. In intercept mode the router intercepts each TCP connection establishment, and determines if the address from which the connection is being initiated is reachable. If the host is reachable, the router allows the connection to be established; otherwise, it prevents the connection.
East(config)# ip tcp intercept list 107 East(config)# access-list 107 permit tcp any East(config)# access-list 107 deny ip any any log East(config)# interface eth0 East(config-if)# description External ethernet interface to net East(config-if)# ip access-group 107 in East(config-if)# exit

TCP intercept is a very effective mechanism for protecting hosts on a network from outside TCP SYN attacks, for extensive details consult the Cisco IOS 12 Security Configuration Guide [5]. The TCP intercept feature is available in most, but not all, Cisco IOS version 12.0 and later releases. Note that TCP intercept, while it can be very useful, can also impose significant overhead on router operations. Examine and test the performance burden imposed by TCP intercept before using it on an operational network.

Land Attack
The Land Attack involves sending a packet to the router with the same IP address in the source and destination address fields and with the same port number in the source


Version 1.1c

Advanced Security Services

port and destination port fields. This attack may cause denial of service or degrade the performance of the router. The example below shows how to prevent this attack.
East(config)# access-list 103 deny ip host host log East(config)# access-list 103 permit ip any any East(config)# interface eth0 East(config-if)# description External interface to East(config-if)# ip address East(config-if)# ip access-group 103 in East(config-if)# exit

Smurf Attack
The Smurf Attack involves sending a large amount of ICMP Echo packets to a subnet's broadcast address with a spoofed source IP address from that subnet. If a router is positioned to forward broadcast requests to other routers on the protected network, then the router should be configured to prevent this forwarding from occurring. This blocking can be achieved by denying any packets destined for broadcast addresses. The example statements below block all IP traffic from any outside host to the possible broadcast addresses ( and for the subnet.
East(config)# access-list 110 deny ip any host log East(config)# access-list 110 deny ip any host log East(config)# interface interface eth0 East(config-if)# ip access-group 110 in East(config-if)# exit

ICMP Message Types and Traceroute
There are a variety of ICMP message types. Some are associated with programs. For example, the ping program works with message types Echo and Echo Reply. Others are used for network management and are automatically generated and interpreted by network devices. For inbound ICMP traffic, block the message types Echo and Redirect. With Echo packets an attacker can create a map of the subnets and hosts behind the router. Also, he can perform a denial of service attack by flooding the router or internal hosts with Echo packets. With ICMP Redirect packets the attacker can cause changes to a host’s routing tables. Otherwise, the other ICMP message types should be allowed inbound. See the example below for inbound ICMP traffic.
East(config)# East(config)# East(config)# East(config)# access-list access-list access-list access-list 100 100 100 100 deny icmp any any echo log deny icmp any any redirect log deny icmp any any mask-request log permit icmp any

For outbound ICMP traffic, one should allow the message types Echo, Parameter Problem, Packet Too Big, and Source Quench and block all other message types. With Echo packets users will be able to ping external hosts. Parameter Problem packets and Source Quench packets improve connections by informing about problems with packet headers and by slowing down traffic when it is necessary. Packet Too Big is necessary for Path MTU discovery. The example below shows a set of filter rules for outbound ICMP traffic that permit these message types.

Version 1.1c


Router Security Configuration Guide

East(config)# East(config)# East(config)# East(config)# East(config)#

access-list access-list access-list access-list access-list

102 102 102 102 102

permit permit permit permit deny

icmp icmp icmp icmp icmp

any any any any any

any any any any any

echo parameter-problem packet-too-big source-quench log

Another program that deals with certain ICMP message types is traceroute. Traceroute is a utility that prints the IP addresses of the routers that handle a packet as the packet hops along the network from source to destination. On Unix and Linux operating systems, traceroute uses UDP packets and causes routers along the path to generate ICMP message types ‘Time Exceeded’ and ‘Unreachable’. An attacker can use traceroute response to create a map of the subnets and hosts behind the router, just as they could do with ping’s ICMP Echo Reply messages. Therefore, block naïve inbound traceroute by including a rule in the inbound interface access list, as shown in the example below (ports 33400 through 34400 are the UDP ports commonly used for traceroute).
East(config)# access-list 100 deny udp any any range 33400 34400 log

A router may be configured to allow outbound traceroute by adding a rule to the outbound interface access list, as shown in the example below.
East(config)# access-list 102 permit udp any any range 33400 34400 log

Distributed Denial of Service (DDoS) Attacks
Several high-profile DDoS attacks have been observed on the Internet. While routers cannot prevent DDoS attacks in general, it is usually sound security practice to discourage the activities of specific DDoS agents (a.k.a. zombies) by adding access list rules that block their particular ports. The example below shows access list rules for blocking several popular DDoS attack tools. [Note that these rules might also impose a slight impact on normal users, because they block high-numbered ports that legitimate network clients may randomly select. You may choose to apply these rules only when an attack has been detected. Otherwise, they would be applied to traffic in both directions between an trusted network and an untrusted network.]
! the TRINOO DDoS systems access-list 170 deny tcp any any eq 27665 log access-list 170 deny udp any any eq 31335 log access-list 170 deny udp any any eq 27444 log ! the Stacheldraht DDoS system access-list 170 deny tcp any any eq 16660 log access-list 170 deny tcp any any eq 65000 log ! the TrinityV3 system access-list 170 deny tcp any any eq 33270 log access-list 170 deny tcp any any eq 39168 log ! the Subseven DDoS system and some variants access-list 170 deny tcp any any range 6711 6712 log access-list 170 deny tcp any any eq 6776 log access-list 170 deny tcp any any eq 6669 log access-list 170 deny tcp any any eq 2222 log access-list 170 deny tcp any any eq 7000 log


Version 1.1c

Advanced Security Services

The Tribe Flood Network (TFN) DDoS system uses ICMP Echo Reply messages, which are problematic to block because they are the heart of the ping program. Follow the directions in the ICMP sub-section, above, to prevent at least one direction of TFN communication.

4.3.4. Example Configuration File
The configuration file shown below is not a complete configuration file. Rather, it provides an example for using access lists on a Cisco router. The diagram below shows the topology that this file is based on. The security policy implemented with the access lists allows most traffic from the internal network to the external network. The policy restricts most traffic from the external network to the internal network.
Protected network

Other networks
Interface eth0

Interface eth1

hostname East ! interface Ethernet0 description Outside interface to the network ip address ip access-group 100 in ! interface Ethernet1 description Inside interface to the network ip address ip access-group 102 in ! ! access-list 75 applies to hosts allowed to gather SNMP info ! from this router no access-list 75 access-list 75 permit host access-list 75 permit host ! ! access-list 100 applies to traffic from external networks ! to the internal network or to the router no access-list 100 access-list 100 deny ip any log access-list 100 deny ip host host log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any log access-list 100 deny ip any host log access-list 100 deny ip any host log access-list 100 permit tcp any established

Version 1.1c


255 access-list 100 permit ospf 14.0 0.0 0.0.0 eq 23 log access-list 150 deny ip any any log ! snmp-server community N3T-manag3m3nt ro 75 ! line vty 0 4 access-class 150 in password 7 123456789012345678901234 login transport input telnet 94 Version 1.255 any range 33400 34400 log access-list 102 deny tcp any range 0 65535 any range 0 65535 log access-list 102 deny udp any range 0 65535 any range 0 65535 log access-list 102 deny ip any any log ! ! access-list 150 applies to admin access from specific hosts no access-list 150 access-list 150 permit tcp host 14.1c .2.2.250 log access-list 102 permit icmp 14.0.0 any source-quench access-list 102 deny tcp any any range 1 19 log access-list 102 deny tcp any any eq 43 log access-list 102 deny tcp any any eq 93 log access-list 102 deny tcp any any range 135 139 log access-list 102 deny tcp any any eq 445 log access-list 102 deny tcp any any range 512 518 log access-list 102 deny tcp any any eq 540 log access-list 102 permit tcp 14.10 host host gt 1023 access-list 100 deny tcp any range 0 65535 any range 0 65535 log access-list 100 deny udp any range 0 65535 any range 0 65535 log access-list 100 deny ip any any log ! ! access-list 102 applies to traffic from the internal network ! to external networks or to the router itself no access-list 102 access-list 102 deny ip host 14.2.255 gt 1023 access-list 100 deny udp any any eq 2049 log access-list 100 deny udp any any eq 31337 log access-list 100 deny udp any any range 33400 34400 log access-list 100 permit udp any eq 53 14.2.0 0.0 0.255 gt 1023 any eq 53 access-list 102 permit udp host 0.0 any packet-too-big access-list 102 permit icmp any parameter-problem access-list 102 permit icmp 14.0.0 0.0 eq 23 log access-list 150 permit tcp host eq 23 log access-list 150 permit tcp host access-list 100 deny tcp any any range 6000 6063 log access-list 100 deny tcp any any eq 6667 log access-list 100 deny tcp any any range 12345 12346 log access-list 100 deny tcp any any eq 31337 log access-list 100 permit tcp any eq 20 Security Configuration Guide access-list 100 deny icmp any any echo log access-list 100 deny icmp any any redirect log access-list 100 deny icmp any any mask-request log access-list 100 permit icmp any 14.255 any echo access-list 102 permit icmp 0.250 host 14.255 gt 1023 any lt 1024 access-list 102 permit udp host

) Once this facility is enabled. To enable turbo access lists on a router. it may not be necessary to keep CAR rules in place at all times.5. For more information on CAR commands. ordered set of rules for the in-bound (receiving) and out-bound (sending) directions. If you use access lists with six or more rules on high-speed interfaces. the most important task that CAR can perform is to mitigate the paralyzing effects of DoS attacks and flash crowds. The output of the command will show both the rules and some traffic statistics about the rate limiting. use the configuration mode command access-list compiled. and a few simple examples. 4. (If your IOS does not support compiled access lists. eliminate excessive traffic. rate-limit {input | output} [access-group [rate-limit] acl] token-bit-rate burst-normal-size burst-excess-size conform-action action exceed-action action To add a rule to an interface.3. the command will generate a harmless error message.6. as shown in the examples below. To remove a rule.1c 95 . IOS will automatically compile all suitable access lists into fast lookup tables while preserving their matching semantics. Each interface can have a separate. This short section gives an overview of CAR. Once you have enabled turbo access lists. The general syntax for a CAR rule is shown below. consult the “IOS Quality of Service Solutions Command Reference” section of the IOS documentation. By allocating a specific amount of bandwidth to defined traffic aggregates. then compiled ACLs can give improved performance. To view the CAR rules on all the interfaces. Using compiled access control lists can greatly reduce the performance impact of long lists. somewhat simplified. You can use CAR to reserve a portion of a link’s bandwidth for vital traffic. and limit spoofed traffic. and later. Rate Limiting with Committed Access Rate Committed Access Rate (CAR) is a router service that gives administrators some control over the general cross-section of traffic entering and leaving a router. in IOS 12.Advanced Security Services 4. CAR Command Syntax Configuring CAR requires you to apply rate limiting rules to each interface where you enforce constraints on traffic or bandwidth usage. Turbo Access Control Lists Some Cisco router models support compiled access control lists. use the command show interface rate-limit. Version 1. but to be ready to apply them quickly when you detect an attack in progress.3. In the latter case. however. or to limit the amount of bandwidth consumed by a particular kind of attack.1(6). data passing through the router can be manipulated to preserve fragile traffic. simply type the rule in interface configuration mode. you can view statistics about them using the command show access-list compiled. called “Turbo ACLs”. enter it again adding the keyword no to the front. A sample of the output is included in the first example below.

and the traffic burst exccess size. use the input keyword. Rate-limit access lists define aggregates based on IP precedence or MAC addresses. before the aggregate uses up its allocated bandwidth. and may also include fine-grained traffic selection specified with an access control list. If the aggregate definition includes an access-group clause. For a more detailed description of the token bucket model. The first action instructs the router on how to handle packets when the aggregate conforms to bandwidth allocation. [It is also possible to apply CAR rules to packets by QoS header and other criteria. the token bucket parameters. consult [9]. CAR Action Syntax drop transmit continue Action Performed Discard the packet. This model gives you a flexible method to stipulate bounds of traffic behavior for an aggregate. • The last section of a rule consists of the two action specifications. there may be as many as nine possible actions. given in bytes.1c . otherwise the access list should be a standard or extended IP access list. The burst excess size denotes the upper bound or maximum size expected for traffic bursts. the most commonly used four are described below. Transmit or forward the packet. it would simply be the size of a typical message. For simple protocols. the traffic burst normal size (in bytes). it indicates that the aggregate is defined by a rate-limit access list. is generally the size of a typical traffic transaction in a single direction. for packets leaving the router use the output keyword. The token bit rate parameter must be specified in bits per second (bps). If the rule is meant to apply to packets entering the router. such as ICMP or DNS. Depending on your IOS version. and the second how to handle packets when the aggregate exceeds its bandwidth allocation. but that is outside the scope of this brief discussion. • The second part of the rate-limit command is comprised of the three token bucket parameters. if you supply no access-group clause then the rule applies to all traffic. The token bucket model needs three parameters for configuration: the token bit rate. • The aggregate definition section of a rule defines the kind of traffic (or “packet aggregate”) to which the rule applies. 96 Version 1. then the CAR rule will apply only to traffic that is permitted by or matches that access list. The burst normal size. The aggregate definition must include the traffic direction. and must be greater than 8000. The CAR facility uses a token bucket model to allocate or limit bandwidth of traffic. Apply the next rate-limit rule.Router Security Configuration Guide Defining Rules Each rate limit rule is made up of 3 parts: the aggregate definition. and the action specifications. It generally describes the allowed rate for the aggregate.] If the keyword rate-limit appears.

and to limit outgoing ICMP ‘ping’ traffic to less than 1% of the link. conformed 2000 bps. 11699 bytes. 50000 extended limit conformed 12 packets. current burst: 0 bytes last cleared 00:02:32 ago. 24990 bytes. 2500 extended limit conformed 130 packets. action: transmit exceeded 0 packets. action: transmit exceeded 0 packets. exceeded 0 bps North# In this second example. you might want to impose both outbound and inbound rate limiting to protect the vital SMTP traffic. action: continue exceeded 255 packets. 0 bytes. current burst: 0 bytes last cleared 00:01:40 ago. 112000 limit. CAR Examples In the first example. The rest of the link’s bandwidth will be usable by excess SMTP traffic and all other IP traffic. 225000 extended limit conformed 346 packets. In practice. 0 bytes. conformed 0 bps. 2500 limit. current burst: 2434 bytes last cleared 00:02:04 ago. action: continue last packet: 2668ms ago. CAR is being used to throttle a TCP SYN flood attack. 25000 limit. action: drop last packet: 7120ms ago. North(config)# North(config)# North(config)# North(config)# no access-list 160 access-list 160 deny tcp any any established access-list 160 permit tcp any any syn interface eth0/0 Version 1. 12740 bytes.Advanced Security Services CAR Action Syntax set-prec-transmit prec Action Performed Set the IP precedence to prec and transmit or forward the packet. action: drop last packet: 7140ms ago. conformed 0 bps. CAR is used to reserve 10% of a 10Mb Ethernet link for vital outgoing SMTP traffic. exceeded 990 bps matches: all traffic params: 9000000 bps.1c 97 . exceeded 0 bps matches: access-group 131 params: 16000 bps. 27074 bytes. North(config)# no access-list 130 North(config)# access-list 130 permit tcp any any eq smtp North(config)# no access-list 131 North(config)# access-list 131 permit icmp any any echo North(config)# access-list 131 permit icmp any any echo-reply North(config)# interface eth0/0 North(config-if)# rate-limit output access-group 130 1000000 25000 50000 conform-action transmit exceed-action continue North(config-if)# rate-limit output access-group 131 16000 8000 8000 conform-action continue exceed-action drop North(config-if)# rate-limit output 9000000 112000 225000 conform-action transmit exceed-action drop North(config-if)# end North# show interface rate-limit Ethernet0/0 Output matches: access-group 130 params: 1000000 bps.

using the class-map command. Before attempting to configure CPP. identify the classes you wish to handle. The control plane consists of the routing.” “malicious. Then service policies should be created and applied that cause traffic classes destined for the route processor to be accepted. all traffic destined for the control plane of a router must be categorized into network administrator-defined groups or classes (e. discarded. use the class map-name command to define rate-limiting policy for each 98 Version 1. or rate limited.Router Security Configuration Guide North(config-if)# rate-limit input access-group 160 64000 8000 8000 conform-action transmit exceed-action drop North(config-if)# end North# The CAR rule in this example simply discards excessive TCP SYN packets. The management plane consists of traffic for configuring and monitoring router operations.1c . To implement a CPP policy.7. Create access lists that match (permit) the traffic from members of each class. 4. Control Plane Policing (CPP) is a Cisco IOS feature that you can employ to counter resource starvation-based DoS attacks that target the central processor of a router (control plane and management plane). Timely and reliable operation of the management and control planes are essential for maintaining the flow of traffic through the forwarding plane. the “critical. consult [10]. For another example of using CAR to combat a DoS attack. Take care when defining and applying CPP policy -. legitimate traffic would also be affected.) 2. Once you have defined your classes.3. In this case. Control Plane Policing (CPP) Conceptually. (If you have a ‘default’ class. Define a named class map for each of the access lists you created in step 1. 3. router operations can be abstracted into three planes: forwarding. and rough traffic rate limits for each of them. Create a policy map using the policy-map command. and management. Detailed information about CPP may be found in a Cisco white paper [12]. CPP protects the central processor via policies that filter or rate limit traffic directed to the processor.g. 1. do not create an access list for is easy to accidentally restrict the wrong traffic and disrupt management or control plane services. signaling and link management protocols. If you knew the general source of the attack (perhaps an IP address range) then you could make the defense more selective by incorporating the address range into the aggregate definition access list. In the map.” and “default” classes). setting up control plane policing on IOS requires four steps.” “normal. The forwarding plane (also called the “data” plane) forwards user data packets through the router. control.

The example below shows how to configure CPP with three different classes: a trusted class for internal and specific external hosts. When planning your CPP rate limits.255 any North(config)# access-list 151 permit ip 14. Traffic from all other hosts will be rate-limited to 150 packets per second. End with CNTL/Z.12. Define a default rate-limiting policy using the command class class-default.1. Apply your policy map to the control plane using the commands control-plane and service-policy.2. North# config t Enter configuration commands. consider the bandwidth from possibly hostile sites. and the bandwidth required to maintain router operations.0.0. one per line.1.255. Traffic from the malicious host will be dropped entirely.Advanced Security Services named class.0.1c 99 .0. Traffic from hosts in the trusted class will have no rate limits. and a default class for all other addresses. North(config)# ! define ACL for CPP trusted hosts North(config)# access-list 151 permit ip 14.0 0.3. allow any rate North(config-pmap-c)# exit North(config-pmap)# class cpp-malicious North(config-pmap-c)# ! drop all traffic in this class North(config-pmap-c)# police rate 10 pps North(config-pmap-c-police)# conform-action drop North(config-pmap-c-police)# exceed-action drop North(config-pmap-c-police)# exit North(config-pmap-c)# exit North(config-pmap)# class class-default North(config-pmap-c)# ! rate-limit all other traffic North(config-pmap-c)# police rate 150 pps North(config-pmap-c-police)# conform-action transmit North(config-pmap-c-police)# exceed-action drop North(config-pmap-c-police)# exit North(config-pmap-c)# exit North(config-pmap)# exit North(config)# ! apply the policy map for CPP North(config)# control-plane North(config-cp)# service-policy input cpp-policy North(config-cp)# end North# Version 1.255 any North(config)# access-list 151 permit ip host 7.4 any North(config)# ! define a class mapping for trusted host North(config)# class-map match-any cpp-trusted North(config-cmap)# match access-group 151 North(config-cmap)# exit North(config)# ! define a class mapping for the malicious host North(config)# class-map match-any cpp-malicious North(config-cmap)# match access-group 152 North(config-cmap)# exit North(config)# ! define our CPP policy map North(config)# policy-map cpp-policy North(config-pmap)# class cpp-trusted North(config-pmap-c)# ! no action here.255. 4.2.0 0. a malicious class for a known hostile host.20 any North(config)# ! define ACL for known hostile host North(config)# access-list 152 permit ip host 1.

0 . [4] Held. 1999. February 1996. Moskowitz. Y..192. “Network Ingress Filtering: Defeating Denial of Service Attacks which employ IP Source Address Spoofing”. McGraw-Hill. [3] Held.. References [1] Chapman. This text provides valuable information on how to packet filter many of the commonly used services. Telnet.. It includes information on TCP Intercept.16. D. “Address Allocation for Private Internets”. 2000. and Hundley. 172. K. and Hundley.1c . and Senie. 1999 This book includes a good introduction to router security.0. O’Reilly Associates. 1995. This Internet ‘Best Current Practice’ RFC gives a good overview of source address filtering. and a good primer on access lists [5] Cisco IOS Release 12. [6] Ferguson. This is the reference manual and guide for major security features in IOS 12. P. This book offers detailed information about access control lists and many examples of list syntax and usage. use the command show policy-map control-plane. Cisco Press. K.0. [2] Karrenberg. The Internet Assigned Numbers Authority has reserved the following three blocks of the IP address space for private intranets: D.. Elizabeth D.255.255. FTP.31. G. and dynamic access lists. use the command no service-policy command as shown below.255. D. McGraw-Hill.. This RFC describes the IP address allocation for private intranets.3.0 . B. 100 Version 1. etc.0. RFC 1918. Building Internet Firewalls. SMTP.0 Security Configuration Guide. RFC 2827. To remove a CPP policy. Cisco Security Architectures. G. reflexive access lists.168.0 .10. Brent and Zwicky.Router Security Configuration Guide To view the current CPP policy and traffic statistics.255. e.8. 1999. North(config)# control-plane North(config-cp)# no service-policy input cpp-policy North(config-cp)# end North# 4. Cisco Access List Field Guide.. and 192.. and Rekhter.g.

[10] “Using CAR During DOS Attacks”. Cisco Press. Version 1.pdf Walks through a detailed CAR example related to ICMP flooding. Detailed directions for doing RFC 2827-compliant filtering on networks connected to multiple providers. Cisco Systems. Feb 2002. Cisco IOS Access Lists. A detailed guide to access lists. April 2002. P. 2005. Cisco ISP Cisco Systems. RFC 3704. P. [9] “Selecting Burst and Extended Burst Values for Class-based Policing”. available at: http://www. It also lists the Cisco IOS releases that support CPP. O’Reilly Associates. and Savola. available at: http://www. Cisco 101 . Cisco white paper. B. “Ingress Filtering for Multihomed Networks”. [8] Sedayao. ps6642/prod_white_papers_list.html This white paper explains the motivations for CPP and provides detailed instructions on how to configure it. 2001.. and an in-depth treatment of Unicast RPF..html Describes the CAR token bucket model and burst size parameters in some depth. Cisco Tech Note. [11] Baker.Advanced Security Services [7] Greene. available under: all with fully worked-out examples. March 2004. [12] “Deploying Control Plane Policing”. including coverage of using access lists with routing protocols. This detailed Cisco guide for Internet Service Providers includes extensive discussion of routing protocols (especially BGP). gives guidance on how to select usable values. Cisco Tech Note. and Smith. 1st Edition.

its foundation is the Internet Protocol (IP). Use of TCP or UDP is immaterial to routing. This section will not provide an in-depth discussion of this protocol.Router Security Configuration Guide 4. Correct. connecting every router to every other would be prohibitively expensive.”[7] These protocols support routed protocols and are used to maintain routing tables. using IP. In practice. RIP. These routers are responsible for directing each IP packet to its intended destination. The two types of protocols are: • Routed protocols – These are protocols that can be routed by a router. consult [16]. as that is far beyond the scope of this guide. • Routing protocols – “A routing protocol gathers information about available networks and the distance. each router can simply be connected directly to every other router. Today. they use one of two transport-layer protocols built on top of IP: the Transmission Control Protocol (TCP) or the User Datagram Protocol (UDP).”[5] This section will discuss two basic types of protocols. of course. Instead. or cost. In a small network. Routed Protocols The most commonly used routed network protocol suite is the TCP/IP suite. each router maintains a route table with information about how to forward packets to their destination addresses. For a detailed introduction to the concepts of routing. and DECnet. to reach those networks. Instead. BGP. and secure operation of any large IP network depends on the integrity of its route tables. IS-IS. Its growth and popularity can be attributed to IP’s ability to connect different networks regardless of physical environment. with a focus on the latter. IPX. it is the basis for the worldwide Internet. consult [6] for a detailed introduction. Some examples of routed protocols are IP. a connected host anywhere on a network can communicate with any other. and EIGRP. The routed protocol allows the router to correctly interpret the logical network. instead it only needs to know the address of one or a small number of routers. efficient. Some examples of routing protocols are OSPF. Routing and Routing Protocols “A protocol is a formal description of a set of rules and conventions that govern how devices on a network exchange information. For larger networks. which takes place exclusively at the network layer. ARPA sponsored the development of IP over twenty-five years ago under the ARPANET project. host applications almost never use raw IP to communicate. 102 Version 1. and the flexible and open nature of the IP network architecture. AppleTalk. All of the examples in this section are based on the sample network architecture shown in Figure 4-1.1c . Each IP host does not need to know a path through the network to every other host. IP is designed for use on large networks.4.

Uses router update messages from other routers to create routes. Border Gateway Protocol version 4 (BGP-4) is the exterior gateway protocol used for conveying route information between autonomous systems on the Internet. An interior gateway protocol (IGP) is used for exchanging routing information between gateways within an autonomous system. and exterior gateway protocols on backbone routers. An autonomous system is a group of networking components under one administrative domain. Although many different dynamic routing protocols exist. is an example of a distance vector based interior gateway protocol. the router Central is connected to the LAN segment 14. is vendor-defined. 3. they can be divided into two groups: interior and exterior gateway protocols. This method is the most flexible because it can automatically adapt to changes in the network. The first three are IETF standards. Uses a manually entered route to a specific ‘gateway of last resort’ when route is not known by any other routing mechanism. to compare routes and to determine the ‘best’ path to a destination. Each router builds its table based on information from the network and from the network administrators. IS-IS. EIGRP. the Routing Information Protocol. and the last. An exterior gateway protocol (EGP) is used between autonomous systems.9. OSPF. and updates the route table.0/24. Dynamic routing.2. although not universal. 4. and EIGRP. The routing algorithm associated with the particular routing protocol determines the optimal path to a particular destination. Border routers might use either. This method usually takes precedence over any other method of routing.Advanced Security Services Route Tables and Routing Protocols A router’s primary responsibility is to send a packet of data to the intended destination. Routers use four primary mechanisms for building their route tables: 1. or both. each router needs a route table. RIP. that interior gateway protocols are employed on interior routers. OSPF. Routers that depend on a single default gateway usually do not use routing protocols. It is typical. Static routing. The gateways within the autonomous system use the route information conveyed by the IGP messages to direct IP traffic. depending on the contents of the table and its routing algorithm. To accomplish this. As network administrator. This method is most useful for border routers and routers that serve as the sole connection between a small LAN and a large network like the Internet. For example. depending on the network architecture in which they are found. Version 1. Direct connection: Any LAN segment to which the router is directly connected is automatically added to the route table. 2. The router then uses a set of metrics. you can manually instruct a router to use a given route to a particular destination.1c 103 . This section focuses on a small number of widely used routing protocols: RIP. IS-IS is an ISO standard. BGP. Default routing.

For general guidance on routing protocols. or [3]. Each update is a full route table. Distance vector protocol: maintains a complex set of metrics for the distance to other networks. the Enhanced Interior Gateway Routing Protocol. Link state protocol: uses a link speed-based metric to determine paths to other networks.Router Security Configuration Guide Open Shortest Path First.g. IS-IS is suitable for large networks. The table below provides a short comparison. Optional metrics are delay. 4. version 4. This is usually called the rate of convergence. Common routing hazards A question that is often overlooked is “Why do we need to concern ourselves with security of the network?” A better question to ask would be “What kind of damage 104 Version 1.and incorporates some features of link state protocols. the number of routers a packet must traverse to reach its destination. and are sent only when the network configuration changes. is a proprietary Cisco IGP that is sometimes used in all-Cisco networks. Intermediate-System to Intermediate-System. and one which is outside the scope of this guide. and IS-IS. Configuring routing in IP networks can be a very complex task. expense and error. Cisco IOS supports only the cost based metric. OSPF is suitable for large networks. BGP-4 employs route aggregation to support extremely large networks (e. Routers establish and maintain neighbor adjacencies every 10 seconds by default. the Border Gateway Protocol. For example. EIGRP. Updates are sent over TCP connections between specifically identified peers. consult the Cisco IOS documentation. EIGRP is suitable for large networks. A complete link state database is broadcast by a designated router every 10 seconds by default to synchronize neighbor route tables. Table 4-2 – Five Popular IP Routing Protocols RIP Distance vector protocol: maintains a list of distances to other networks measured in hops. are examples of link state interior gateway protocols. partly because the maximum distance is 15 hops. Broadcasts updates every 90 seconds to all EIGRP neighbors. Updates are sent via multicast. Link state protocol: uses a cost-based metric by default to determine paths to other networks. RIP is suitable only for small networks. is the IETF standard exterior gateway protocol. Broadcasts updates every 30 seconds to neighboring RIP routers to maintain integrity. the Internet). Each update includes only changes to the network. Routing does raise several security issues. BGP-4. Each router maintains a simplified map of the entire network. Each update only includes changes to the network.4. this section discusses some of these security issues and describes several of the security services in moderate detail.1c . OSPF IS-IS EIGRP BGP Another important aspect of a routing protocol scheme is the amount of time it takes for network architecture or connectivity changes to be reflected in the route tables of all affected routers. A distance vector exterior gateway protocol that employs a sophisticated series of rules to maintain paths to other networks. and Cisco IOS offers several security services for routing.1. in a large network OSPF offers much faster convergence than RIP.

For example. Authenticate route table updates – By using routing protocols with authentication.Advanced Security Services could an adversary do to our network?” Section 3 presents some motivations for overall router security. ARP is described in more detail in RFC 826 and Parkhurst [2]. or ARP. ARP and LANs Address Resolution Protocol. 2. preventing router update messages from being sent or received will result in bringing down parts of a network. For a detailed description of Proxy ARP. because ARP offers no security. By doing this. To resist denial of service attacks. A detailed analysis of routing protocol threats and countermeasures may be found in a Cisco SAFE white paper [45]. Authenticated router updates ensure that the update messages came from legitimate sources. However. • protect mission information from unauthorized exposure and modification. is the protocol used to map IP addresses to a particular MAC or Ethernet address. Another form of attack an adversary might attempt against a router is a denial of service attack. the attacker can re-route network traffic in whatever manner he desires. The fundamental security weakness of ARP is that it was not designed to use any form of authentication. consult RFC 1027. Routing security should be a top priority for network administrators who want to: • prevent unauthorized access to resources on the network. • prevent network failures and interruptions in service. routers need rapid convergence and backup routes.1c 105 . network administrators can deter attacks based on unauthorized routing changes.4. Anyone on a LAN segment can modify an entry in the ARP cache of Version 1.2. and recover from them quickly. bogus messages are automatically discarded. This can be accomplished in many different ways. 4. but is unsuitable for large networks because it increases administrative overhead and requires administrative response to any failures. an attacker who sends false routing update packets to an unprotected router can easily corrupt its route table. Proxy ARP is a method of routing packets using the Ethernet MAC address instead of the IP address to determine the final destination of a packet. Use only static routes – This may work in small networks. There are two basic approaches available for protecting route table integrity: 1. An unprotected router or routing domain makes an easy target for any network-savvy adversary. This section focuses on security issues related to routing and routing protocols. The key to preventing such an attack is to protect the route tables from unauthorized and malicious changes. For example. neither does Proxy ARP.

or simply giving access to certain packets of data to an unauthorized person. 4. Section 4. when applied correctly. malicious. it signs the 106 Version 1. The receiving router checks the shared secret to determine whether the message should be accepted. That is. Each sending router then uses this key to ‘sign’ each route table update message.1c . Central(config)# interface ethernet0/0 Central(config-if)# no ip proxy-arp Central(config-if)# exit Central(config)# interface ethernet0/1 Central(config-if)# no ip proxy-arp Central(config-if)# end Central# End with CNTL/Z. and routing protocols This section describes how to protect routers from some common routing hazards. In any case. a denial of service. or corrupted routing updates that would compromise the security or availability of a network. authentication occurs when two neighboring routers exchange routing information. routers connected to the same network segment all use a shared secret key. Authentication ensures that the receiving router incorporates into its tables only the route information that the trusted sending router really intended to send.4. and it should be disabled. authentication in RIP version 2 and EIGRP work in a similar fashion. This sub-section describes the implementation of router neighbor authentication in OSPF. BGP-4. Plaintext authentication uses a shared secret key known to all the routers on the network segment. Therefore. OSPF Authentication Router neighbor authentication is a mechanism that. OSPF uses two types of neighbor authentication: plaintext and message digest (MD5).5 gives some security guidance for one exterior gateway protocol. It prevents a legitimate router from accepting and then employing unauthorized. Central# config t Enter configuration commands. if a host on the network does not use default gateways.3. In this case.Router Security Configuration Guide a router that serves the segment. The following example illustrates how to do just that. Such a compromise might lead to re-routing of traffic. static routes. Route tables. can prevent many routing attacks. one per line. The main focus of this section is using peer router authentication with interior gateway protocols. it is susceptible to bad or malicious routes. Router Neighbor Authentication The primary purpose of router neighbor authentication is to protect the integrity of a routing domain. but instead uses Proxy ARP to handle the routing. because it is a good illustration of the basic principle. Each router accomplishes authentication by the possession of an authentication key. When a sending router builds an OSPF packet. Proxy ARP is generally not used anymore.4.

routers North.0/24 Figure 4-3: An Example Routing Architecture In this example. the router rejects the packet. Internet eth0/0 North eth0/1 14.10.64/24 14.9.0/24 eth0/0 14. and Central all share the same secret key. This method does not provide much security because the key is in plaintext in the packet. The second.64/24 South eth0/1 14.1c 107 .2.Advanced Security Services packet by placing the key as plaintext in the OSPF header.0/16 eth1 eth0 Figure 4-3 shows our example network with its routing protocols.1. If the keys match. r0utes-4-all.250/24 RIP Second Floor 14.9. Using this method reveals the secret key to any attacker using a network sniffer on the right LAN segments.250/16 14.250/24 eth0/0 14.2.10. whose cryptographic Version 1.0. they can pose as a trusted router. East.0/24 Central eth0/1 Autonomous System Border Router (ASBR) 14. The receiving router then compares the received key against the key in memory.6. Each of these routers authenticates to each other using the MD5 message digest authentication method. Otherwise.2. then the router accepts the packet.1. Once an attacker captures the key.2.1.9. is message digest authentication.1. with a Key ID of 1.20/16 East 14.250/16 OSPF Area 0 Facility Network 14.6.2. and more secure method.2.

is placed in the header. it rejects the packet as being invalid. both header and body. The receiving router. looks at the Key ID to determine which key was used to generate the hash. or signature. 1. then the North trusts the packet. the secret key is appended to the packet. The sequence number is incremented with every new packet. Otherwise. The output. the key is r0utes-4-all. North. If the regenerated hash matches the hash that was sent from East. MD5. use the superior MD5 method. East runs the cryptographic hash algorithm. Figure 4-4 shows how East authenticates to North. The router then uses its own key to regenerate the hash on the received packet in the same manner as the sending router. is written over the secret that was appended to the packet.1c . This sequence number protects against replay attacks so that no two OSPF packets will have the same hash value. OSPF Version OSPF pkt type OSPF Version OSPF pkt type OSPF length Source OSPF Router ID OSPF length Source OSPF Router ID OSPF Area ID 0 (no checksum) 2 (cryptographic auth type) 0 1 (Key ID) 16 (MD5 len) MD5 hash algorithm OSPF Area ID 0 (no checksum) 2 (cryptographic auth type) 0 1 (Key ID) 16 (MD5 len) Cryptographic sequence number OSPF packet body Cryptographic sequence number OSPF packet body 16-byte secret MD5 cryptographic hash Figure 4-4: OSPF Calculation of an MD5 Authentication Signature (from [4]) OSPF Plaintext Authentication This method is not recommended. In this case. against the OSPF packet. It then picks a primary key to use on the network segment. Finally. East also places a 32-bit sequence number in the header.Router Security Configuration Guide authentication type is denoted by a value of 2. East first builds an OSPF packet. The corresponding Key ID. 108 Version 1. 16 bytes.

router Central would also have to be configured with MD5 authentication and the same shared key as shown below.1. use the superior MD5 method. North# config t Enter configuration commands. North(config)# router ospf 1 North(config-router)# network 14.255 area 0 East(config-router)# exit East(config)# int eth0 East(config-if)# ip ospf message-digest-key 1 md5 r0utes-4-all East(config-if)# end East# RIP Authentication The RIP routing protocol also supports authentication to prevent routing attacks.Advanced Security Services OSPF MD5 Authentication The example below illustrates an example of setting up MD5 for OSPF router neighbor authentication. RIP Plaintext Authentication This method is not recommended. Using the example network shown in Figure 4-1. Central and South. The example transcripts below show routers North and East receiving the key r0utes-4-all. End with CNTL/Z.6. below.255.0. The example transcripts below show routers from Figure 4-3.0.255 area 0 North(config-router)# area 0 authentication message-digest North(config-router)# exit North(config)# int eth0/1 North(config-if)# ip ospf message-digest-key 1 md5 r0utes-4-all North(config-if)# end North# East# config t Enter configuration commands. Each sending router uses these keys to generate the cryptographic hash incorporated into each RIP update message.0 0. all the routers participating in a given network should be configured in the same way.255 area 0 East(config-router)# network 14.1c 109 . The receiving router then uses the shared secret to check the hash and determine whether the message should be accepted.0 0. East(config)# router ospf 1 East(config-router)# area 0 authentication message-digest East(config-router)# network 14. In practice. contained in their respective Version 1. RIP’s method of authentication is very similar to that of OSPF. one per line.255. RIP MD5 Authentication The example below illustrates an example of setting up MD5 for RIP router neighbor authentication. End with CNTL/Z. using the same key. one per line.2.0.1. although the IOS commands are somewhat different. receiving the key my-supersecret-key.0.0 0. The neighboring RIP routers use shared secret keys.0.0.

However. A key chain is a container that holds multiple keys with the associated key IDs and key lifetimes. only one authentication packet is sent.Router Security Configuration Guide key chains. Central(config)# router rip Central(config-router)# version 2 Central(config-router)# network 14. End with CNTL/Z. one per line. one per line. and uses the first valid key that is encountered. In practice.0. The other key is usually used when migrating to different keys. one per line. End with CNTL/Z. all the routes connected to a given network must be configured in the same way. Central# config t Enter configuration commands. The example below shows how to enable version 2 of RIP. That is. RIP manages authentication keys by the use of key chains.0. all of them must possess the same shared key(s). Multiple keys with different lifetimes can exist. South(config)# key chain SOUTH-KC South(config-keychain)# key 1 South(config-keychain-key)# key-string my-supersecret-key South(config-keychain-key)# exit South(config-keychain)# key 2 South(config-keychain-key)# key-string my-othersecret-key South(config-keychain-key)# end South# RIP version 1 did not support authentication. Prior to enabling RIP MD5 authentication. both routers will only use the first valid key. This was a feature that was included in RIP version 2. End with CNTL/Z. Central(config)# key chain CENTRAL-KC Central(config-keychain)# key 1 Central(config-keychain-key)# key-string my-supersecret-key Central(config-keychain-key)# exit Central(config-keychain)# key 2 Central(config-keychain-key)# key-string my-othersecret-key Central(config-keychain-key)# end Central# South# config t Enter configuration commands. The router examines the key numbers in order from lowest to highest. Each RIP router must first be configured to use version 2 in order to enable authentication during routing updates. End with CNTL/Z. each neighboring router must have a shared secret key. However.0 Central(config-router)# end Central# South# config t Enter configuration commands. In the example below. one per line.1c . 110 Version 1. Both key chains share the keys my-supersecret-key and my-othersecretkey. Central and South have key chains named CENTRAL-KC and SOUTH-KC. Central# config t Enter configuration commands.

0. End with CNTL/Z.Advanced Security Services South(config)# router South(config-router)# South(config-router)# South(config-router)# South# rip version 2 network 14. North and East. the example below shows how to enable authentication for RIP. key number. MD5 authentication is enabled within autonomous system 100 and linked to a particular key chain. Authentication for RIP is enabled on the interfaces. EIGRP is configured on both routers for the 14. Router North has associated key number 1 with the Version 1. Enable authentication for EIGRP messages. Configure key management (optional). and key string to be used. Central will be using the key chain CENTRAL-KC that was created earlier and the MD5 method of authentication. 4. Initially. the key chain is defined within key chain configuration mode consisting of a key name.1. Central# config t Enter configuration commands.0. This insures the integrity of routing messages accepted from neighboring routers. 3. To configure EIGRP authentication: 1. Central(config)# int ethernet0/1 Central(config-if)# ip rip authentication key-chain CENTRAL-KC Central(config-if)# ip rip authentication mode md5 Central(config-if)# end Central# South# config t Enter configuration commands. In the example below. key number. South(config)# int ethernet0/0 South(config-if)# ip rip authentication key-chain SOUTH-KC South(config-if)# ip rip authentication mode md5 South(config-if)# end South# EIGRP Authentication EIGRP route authentication is provided through the use of a keyed Message Digest 5 (MD5) hash. Select the MD5 authentication mode. one per line. Specify the key chain. End with CNTL/Z.0.0 end Finally.0/16 network. Router North’s key chain is defined as northkc and router East’s key chain is named eastkc. Proceeding into the interface configuration mode. one per line. Finally. In this example. The example below details the steps necessary to configure MD5 authentication on two EIGRP peers. 2. and key string. The key chain name is locally significant and neighboring routers do not have to be configured with the same name.1c 111 .

0. In the example above. one per line. End with CNTL/Z.0 255. but only one authentication packet is sent.0 255.255. In this case.0 North(config-router)# exit North(config)# interface eth 0/1 North(config-if)# ip authentication mode eigrp 100 md5 North(config-if)# ip authentication key-chain eigrp 100 NORTH-KC North(config-if)# exit North(config)# key chain NORTH-KC North(config-keychain)# key 1 North(config-keychain-key)# key-string my-secret-key North(config-keychain-key)# send-lifetime 00:00:00 Oct 1 2003 00:00:00 Jan 1 2004 North(config-keychain-key)# accept-lifetime 00:00:00 Oct 1 2003 00:00:00 Jan 7 2004 North(config-keychain-key)# end North# East# config t Enter configuration commands.End with CNTL/Z.0.2. 2004. the routers will send updates authenticated with the key ‘my-secret-key’ from October 1.0. Key management is optionally configured with the acceptlifetime and send-lifetime commands.Router Security Configuration Guide key-string ‘secret-key’.1. it will accept updates with that key until January 7. the key-string “secret-key” is associated with key number 1.255. 2004.255. East(config)# router eigrp 100 East(config-router)# network 14.0.0 East(config-router)# passive-interface eth1 East(config-router)# exit East(config)# interface eth 0 East(config-if)# ip authentication mode eigrp 100 md5 East(config-if)# ip authentication key-chain eigrp 100 EAST-KC East(config-if)# exit East(config)# key chain EAST-KC East(config-keychain)# key 1 East(config-keychain-key)# key-string my-secret-key East(config-keychain-key)# send-lifetime 00:00:00 Oct 1 2003 00:00:00 Jan 1 2004 East(config-keychain-key)# accept-lifetime 00:00:00 Oct 1 2003 00:00:00 Jan 7 2004 East(config-keychain-key)# end East# It is important to note that each key string is associated with a specific key number.255. North# config t Enter configuration commands.1. North(config)# router eigrp 100 North(config-router)# network 14.6.0 255. 2003 until January 1. 112 Version 1. Multiple keys and key-strings can be configured on a router. The examples below show how to configure EIGRP authentication and keys.0 East(config-router)# network 14.1c . one per line. The router chooses the first valid key while examining the key numbers from lowest to highest.

one per line. The transcripts below show routers from Figure 4-5. they can pose as a trusted router. and 3. HMAC-MD5 authentication. Using this method reveals the secret key to any attacker using a network sniffer on the associated LAN segments. Serial Hello CNSP and PSNP. North and East.1c 113 . in their respective key chains. Enhanced Clear Text authentication functions like Plaintext authentication with the exception that the authenticating key is encrypted only within the configuration of the router. The IS-IS HMAC-MD5 authentication feature adds an HMAC-MD5 digest to each IS-IS protocol data unit (PDU) packet before transmitting the packet across the network. an IS-IS routing protocol instance. LAN Hello. being configured with the key my-secret-key.” [26] In order to use IS-IS HMAC-MD5 authentication. The receiving router then uses its identical set of keys to check the IS-IS authentication. but the key itself is not sent. Enhanced Clear Text and Hashed Message Authentication Code Message Digest 5 (HMAC-MD5). The authentication can be enabled on the two different IS-IS Levels (Level 1 and Level 2) independently. preventing it from being read while it is being transmitted. North# config t Enter configuration commands. End with CNTL/Z. “The interface-related PDUs (LAN Hello. The message digest is created using a shared secret key and a message. 2. The IS-IS HMAC-MD5 authentication can be applied to all five types of PDU. North(config)# key chain ISIS-KC North(config-keychain)# key 1 North(config-keychain-key)# key-string my-secret-key North(config-keychain-key)# end North# Version 1. a key chain. you must configure the following three things: 1. This method does not provide much security because the key is in plaintext in the packet. This process prevents unauthorized PDUs from being accepted and used in the IS-IS routing domain. Each sending router uses its keys to generate the cryptographic hash incorporated into IS-IS messages. and recommended security practice is to use HMAC-MD5 authentication.Advanced Security Services IS-IS Authentication IS-IS provides three methods of authentication to prevent routing attacks: Plaintext (or clear text). with different levels and different passwords. IS-IS has five PDU packet types: LSP. The digest is used by the receiving router(s) to authenticate each PDU. The keys on the key chain must be identical on neighboring routers. Serial Hello. CNSP and PSNP) can be enabled with authentication on different interfaces. The final. This method sends a “message digest” instead of the authenticating key itself. Plaintext authentication uses a shared secret key known to all the routers on the network segment. Once an attacker captures the key.

1. The following scripts implement this security policy. Notice that IS-IS HMAC-MD5 authentication is applied to the North router’s ethernet interface 0/1 and on the East router’s ethernet interface 0 to authenticate neighbor adjacencies (i.0000.0000.2.1.e. End with CNTL/Z.6.00 North(config-router)# is-type level-1 North(config-router)# authentication mode md5 level-1 North(config-router)# authentication key-chain ISIS-KC level-1 North(config-router)# exit North(config)# interface ethernet 0/1 North(config-if)# ip address 14. and PSNP packets): North# config t Enter configuration commands.1. Also notice that IS-IS HMAC-MD5 authentication is applied to the IS-IS instance to authenticate routing database updates (i. Level 1: LSP.1.0001. East(config)# key chain ISIS-KC East(config-keychain)# key 1 East(config-keychain-key)# key-string my-secret-key East(config-keychain-key)# end East# Internet eth0/0 North eth0/1 NET = 49.0003.1.0001.0000.1c .0.0/24 Central Figure 4-5: An Example Routing Architecture Next.2.250/24 eth0/0 14.250 255.0000.0 North(config-if)# ip router isis secure-network 114 Version 1.1. LAN Level 1: Hello packets).1.250/16 NET = 49. For example. End with CNTL/Z.0001. a site’s security policy may require that routers authenticate to establish neighbor adjacencies within an area and perhaps it requires that routers authenticate before accepting routing database updates within an area. one per line. ensure the IS-IS routing protocol process is running on the routers.15.0001.1. Then select the IS-IS PDU types to authenticate and enable IS-IS HMAC-MD5 authentication on the routers.0000. one per line.0001.20/16 East 14.250/16 14.0.6.Router Security Configuration Guide East# config t Enter configuration commands.0/16 14.255.00 14.0001 Facility Network 14.e. CNSP.00 eth1 eth0 IS-IS Area 49. North(config)# router isis secure-network North(config-router)# net 49.0000.

0000.0001.” [27] Key Management The strength of these methods. Version 1. Therefore an attacker has to be physically attached to a router in the IS-IS network to maliciously disrupt the IS-IS routing environment. However. The other issue with maintaining secrecy is the question of “How many keys should be used in the routing domain?” That is. which makes them susceptible to remote access by intrusive applications. Using a separate key for each router neighbor-to-neighbor connection can become an administrative nightmare. OSPF uses single keys. or a separate key for each router neighbor-to-neighbor connection. Key lifetime is also important. End with CNTL/Z.0003. IS-IS offers a unique security advantage compared to other IP routing protocols. The best method for distributing keys to trusted routers is to do it manually.1. such as RIP. OSPF and BGP. IS-IS.1. “Other routing protocols.Advanced Security Services North(config-if)# isis authentication mode md5 level-1 North(config-if)# isis authentication key-chain ISIS-KC level-1 North(config-if)# end North# East# config t Enter configuration commands. because failure to do so can compromise the entire network.00 East(config-router)# is-type level-1 East(config-router)# authentication mode md5 level-1 East(config-router)# authentication key-chain ISIS-KC level-1 East(config-router)# exit East(config)# interface ethernet 0 East(config-if)# ip address 14. and EIGRP use Cisco IOS key chains. IS-IS. depends on two factors: the secrecy of the keys and the quality of the keys. and EIGRP routing update authentication.0 East(config-if)# ip router isis secure-network East(config-if)# isis authentication mode md5 level-1 East(config-if)# isis authentication key-chain ISIS-KC level-1 East(config-if)# end East# Authentication on Level 2 packets is accomplished by changing the level-1 parameter in the example above to level-2.20 255. which offer substantial control over key lifetime. East(config)# router isis secure-network East(config-router)# net 49. RIP. RIP. OSPF. one per line.1c 115 . maintaining the secrecy of the key becomes much more important.0. so using a common key for the entire routing domain is recommended.255. are susceptible to attacks from remote IP networks through the Internet because routing protocol packets are ultimately embedded in IP packets.0000. A key’s secrecy is intact only if it is known by the trusted routers but hidden from any attacker. whether one key should be used for the entire routing domain. an administrator must manually change the keys when their lifetimes expire. ISIS packets are encapsulated over the data link and are not carried in IP packets.

2). Using static routes exclusively can make network administration very difficult. However. Central(config)# ip route 14. If you use routing update authentication.5. These routes typically take precedence over routes chosen by dynamic routing protocols. such as OSPF. The router must be configured to maintain correct time (see Section 4. and RIPv2 authentication.1c . The number of different keys and the key validity periods should be defined in the router security policy. Static Routes Static routes are manually configured on the router as the only path to a given destination. See Section 4. The following example illustrates how to create a static route with a higher administrative distance than OSPF. EIGRP.255.6. static routes are very secure.1 for a detailed description. OSPF-derived routes have a default administrative distance of 110. In most cases. one per line. can correctly re-route traffic in the case of a router or link failure. if an administrative distance is specified.20 120 Central(config)# end Central# 116 Version 1. Static routes have a default administrative distance of 1. such as link failures.1. Also. Central# config t Enter configuration commands.0 14. The rules for generating good passwords apply to generating good keys as well. configuring a large network to use only static routes can make the availability of large pieces of the network subject to single points of failure. For example.255. then your router security policy should define the key management procedures and responsibilities. Static routes cannot easily handle events that change the network topology.1. then that static route can be overridden by dynamic information. For more information on the internal workings of static routes. Both of these are configured within the key chain configuration mode. but not OSPF authentication. The other factor that authentication relies upon is the quality of the keys. and specify the start-time and end-time to accept and send individual keys. In one sense. End with CNTL/Z. so they can be used for IS-IS.2. consult [7].0 255. static routes take precedence over their dynamic counterparts. They are not vulnerable to spoofing attacks because they do not deal with router update packets.Router Security Configuration Guide Management of key lifetime is accomplished optionally through the use of the keychain accept-lifetime and send-lifetime commands. A dynamic routing protocol. Thus a static route must have an administrative distance greater than 110 if the OSPF derived route is to have precedence over the static route. These commands apply to keys in a keychain.

Hello 10. which is 40 seconds in the example. constantly sending routing updates will likely expose the identity of the infiltrator. Central# show ip ospf interface ethernet0/0 Ethernet0/0 is up. . .1c 117 . Reducing both of these timers causes routing to switch to an alternate path more quickly in the event of a failure. then Central starts an SPF calculation after receiving a topology change.150/24. . reducing convergence time. then reducing the convergence time will cause that false route to die quickly. as discussed in Section 4. Retransmit 5 Hello due in 00:00:05 .4. . State DROTHER. The output of the show ip ospf interface interface command shows that the time between Hello packets on interface ethernet0/0 is 10 seconds. It will also wait 10 seconds between two consecutive SPF calculations.168.Advanced Security Services The simplest approach for discarding traffic with black-hole routes is to set up static routes. The timers for OSPF can be viewed by using the show ip ospf pid command and the show ip ospf interface interface command. The output of the show ip ospf pid command shows that OSPF on Central will perform an SPF (Shortest Path First) calculation 5 seconds after it receives a topology change. Hold time between two SPFs 10 secs . line protocol is up Internet Address 192. is the time hello packets must not have Version 1. Convergence Reducing the convergence time (the time it takes for all routers to learn of a change in the network) can improve the level of security. Dead 40. then two consecutive SPF calculations can be done without any waiting period. If an attacker creates a spoofed route to redirect traffic. Wait 40. will increase network load. The example below illustrates how to reduce convergence on an OSPF and RIP network. Priority 1 . If this value is 0. As a cautionary note.20. The Dead interval. However. In either case. Timer intervals configured. unless the attacker continues to send routing updates. different aspects of network security will be addressed. SPF schedule delay 5 secs. Area 1 Transmit Delay is 1 sec. If this value is 0. especially when using RIP. The default settings have been selected to provide optimal performance.6. Central# show ip ospf 1 .

If the Hello-interval and Dead-interval are modified on a router. Priority 1 . Wait 20. The second modification sets the Hello-interval to 5 seconds. . all routers on that network must have the same Hello-interval and Dead-interval. End with CNTL/Z. then all other OSPF routers on that network must be changed as well. The first modification sets the SPF calculation delay to 1 second and the delay between two consecutive SPF calculations to 4 seconds. 118 Version 1.20. That is. the Retransmit-interval to 8 seconds. State DR. Hold time between two SPFs 4 secs .168.150/24. one per line. The Retransmit interval is the time between LSA (Link State Advertisement packets sent by OSPF) retransmissions. . Central(config)# interface ethernet0/0 Central(config-if)# ip ospf hello-interval 5 Central(config-if)# ip ospf dead-interval 20 Central(config-if)# ip ospf retransmit-interval 8 Central(config-if)# ip ospf transmit-delay 6 Central(config-if)# end Central# show ip ospf interface ethernet0/0 Ethernet0/0 is up. must be greater than the expected round trip between Central and any other router on the same network. the Dead-interval to 20 seconds. line protocol is up Internet Address 192. Central# show ip protocols . Central# config t Enter configuration commands. Dead 20. and the Transmit-delay to 6 seconds. Similarly. SPF schedule delay 1 secs. Hello 5. Retransmit 8 Hello due in 00:00:02 . the timers for RIP can be viewed by using the show ip protocols command. .Router Security Configuration Guide been seen before Central declares its neighbor dead.1c . Otherwise. The Transmit Delay is the time in seconds that Central will take to transmit a link-state update packet. the routers will be sending needless LSA packets. which is 5 seconds. Timer intervals configured. This time. Central# config t Central(config)# router ospf 1 Central(config-router)# timers spf 1 4 Central(config-router)# end Central# show ip ospf . The example below shows how to modify OSPF timers. . Area 1 Transmit Delay is 6 sec.

flushed after 230 . an update will be delayed before transmission. is the amount of time. .255. End with CNTL/Z. which is not shown. 4.0. flushed after 240 . Router1# show config . next due in 22 seconds Invalid after 180 seconds.4. The flush time is the amount of time that a route will remain in the routing table before it is removed if no update to that route is received. consult Chapter 13 of [2]. . Routing Protocol is "rip" Sending updates every 30 seconds. The example below illustrates such a case. one per line.Advanced Security Services . It can also be used to keep any unnecessary parties from learning about the existence of certain routes or routing protocols used. . and this is preferable to running RIP on an entire large network. Central# In general. For details on this topic. hold down 150. Disabling unneeded routing-related services Passive Interfaces The passive-interface command is used to prevent other routers on the network from learning about routes dynamically.1c 119 .0. The hold down time is the time that a route will remain in the routing table before a new route is accepted.1. If no update is received within 180 seconds. Routing Protocol is "rip" Sending updates every 20 seconds. In its current configuration. then the route is declared invalid. RIP routing updates are sent every 30 seconds.15. measured in milliseconds. interface ethernet0 description Active routing interface for 14.1. The sleep time. The example shows how to modify the RIP timers.0 ! Version 1.4. Central# config t Enter configuration commands.0 net ip address 14. Central(config)# router rip Central(config-router)# timers basic 20 120 150 230 3 Central(config-router)# end Central# show ip protocols .250 255. OSPF and IS-IS are preferable to RIP. . It is typically used when the wildcard specification on the network router configuration command configures more interfaces than desirable. hold down 180. It is also possible to redistribute OSPF or IS-IS routes over RIP. next due in 6 seconds Invalid after 120 seconds. ! router ospf 1 network 14.0. .0. Router1# show config .0. OSPF has been enabled to run on all subnets of 14.0 net ip address enabling RIP on Central will cause RIP broadcasts to be sent out of interfaces ethernet0/0 and ethernet0/1.0 net ip address 14. However.150 255. .1. Thus. this command blocks routing updates from being sent or received on an interface.255. although ethernet0/0 is part of an OSPF network.2.250 255. OSPF will run only on interfaces ethernet0 and ethernet1. In Figure 4-3.Router Security Configuration Guide interface ethernet1 description Active routing interface for 14.0. RIP broadcasts will be sent through that interface.255.3. When used on RIP. as shown below.0.x.0. Router1# When used on OSPF.150 255.0 ! interface ethernet1 ip address 14.0. The reason for this is that both interfaces appear to have the same Class A internet address.50 255. The example below illustrates how to remedy that problem.0.e. interface ethernet0 ip address because that version only uses major network numbers.0.2.255 area 0 passive-interface ethernet2 . this command stops routing updates from being sent out on an interface. 14.0 0.255.50 255.x.0 0. i.90.0.255 area 0 .3.1. Router1# This command functions slightly differently on RIP.13.0 0.1c . This command is especially important when using RIP version 1.2. An alternative method to this is to simply not enable OSPF on certain interfaces.15. 120 Version 1.255 area 0 network 14. . but routing updates will still be received and processed. In the example above. by designating ethernet2 as a passive interface.0 ! interface ethernet2 ip address 14.0 ! router ospf 1 network 14.0 ! interface ethernet2 description Passive interface on the 14.x.0.90.0.

To filter the receipt of network updates. Thus all downstream routers will learn about the networks that were supposed to be filtered in these LSAs.2.0. Central(config)# router rip Central(config-router)# passive-interface ethernet0/0 Central(config-router)# end Central# The syntax for using this command on OSPF is nearly identical.0/24 will be used.10. To apply this command to a routing protocol.2. For illustration purposes. End with CNTL/Z.3. Each command behaves differently with respect to the routing protocol used. Central(config)# router ospf 1 Central(config-router)# passive-interface ethernet0/1 Central(config-router)# end Central# Using filters to block routing updates The distribute-list command is used to apply access lists on routing protocols. routes from a different autonomous system (AS).255 Central(config)# access-list 55 permit any Central(config)# end Central# The OSPF distribute-list in configuration command prevents routes from being inserted into the routing table. To suppress advertisements of particular networks in updates. Central# config t Enter configuration commands. one per line. you must first create an access list.0. but it does not stop routes from being sent out in the link-state advertisements (LSAs).2. an access list with rules filtering out 14. The following example shows how to prevent Central from advertising the 14. However. including Parkhurst [2]. Some authors.10. End with CNTL/Z. the following example is for illustration purposes only. one per line. see Section 4. This command has two primary functions. For more information about how to create access lists.10.1c 121 . use the distribute-list out command . With this setting North and East would not see a route to the 14. End with CNTL/Z. since OSPF is not enabled on the interface to the RIP network.Advanced Security Services Central# config t Enter configuration commands. Version 1. Central# config t Enter configuration commands.0 network from the RIP routing domain into the OSPF routing domain. The distribute-list out command in OSPF configuration mode stops routes from being advertised in updates.2. Therefore. use the distribute-list in command. that is. this restriction only applies to external routes. however. advise against using distribute-list in for OSPF.10. The example below illustrates that.0 0. Central(config)# access-list 55 deny 14.0 network. one per line. this step is unnecessary.

Connections from hosts on the 14.0 network are prevented from reaching the Internet.255 South(config)# access-list 55 permit any South(config)# router rip South(config-router)# distribute-list 55 out South(config-router)# end South# The examples above essentially accomplish the same task. South# config t Enter configuration commands. East and Central will not see routes to the 14.0.10. South(config)# access-list 55 deny 14. North and East will not see a route to 14. Central(config)# router ospf 1 Central(config-router)# distribute-list 55 out Central(config-router)# end Central# The RIP distribute-list in command deletes routes from incoming RIP updates. Central still has the route in its table.2. but that is a drastic approach.0.0 network as it comes in from a RIP update from South. The following example shows Central deleting the route to 14. Central# config t Enter configuration commands.) 122 Version 1.0 network can communicate with hosts on the 14.2. the easiest way to prevent hosts on the 14.10. it will not advertise this network to other routers. as their default gateway.1c . hosts on the 14. the effect of applying the same filter used in the previous examples to South is that North.Router Security Configuration Guide Central# config t Enter configuration commands. End with CNTL/Z.0 network if the hosts on the latter network use Central.0 network. as their default gateway.2. However.10. that is.2.10. one per line.0 network. The second and third filters each fix the problem that was evident with the first filter. (Of course. Thus.10.0. However. while Central is not advertising a route to the 14.2.10. since Central no longer has a route to network 14.0 This is because. Subsequently.10. End with CNTL/Z.2. This is because either Central is filtering the routes it receives (second filter) or South filters the routes it advertises (third filter). Using the first filter. hosts from the 14.1. End with CNTL/Z. network if the hosts on the latter network use South. South still maintains a route to the 14. thereby preventing North from learning that route.2. one per line. Therefore.0. Thus.0 network from communicating with hosts on any other subnets is to simply turn off interface Ethernet0/1 on South. instead of Central. one per line. Central(config)# router rip Central(config-router)# distribute-list 55 in Central(config-router)# end Central# The RIP distribute-list out command prevents routes from being advertised in updates. In either case.0 network can be made with hosts on the 14. instead of North.0 network because it is directly connected to it. the three different filters also have unusual side effects.10. all updates sent from that router will not advertise the deleted route. a similar problem arises.

OSPF is the superior routing protocol. prefix list filtering. and route flap damping. see [2] for detailed directions.7.5. consult [15]. but an attacker could then take advantage of RIP and insert a malicious route into the routing table. check the configured protocols using the command show ip protocols. While both protocols support authentication. then migrating to OSPF is the recommended solution. but OSPF offers better scaling and faster convergence. Exterior Gateway Routing Protocol Security Configuring an exterior gateway protocol can be very complex. and [30] through [43]. is part of AS 27701. [17]. Failure to do so will not cause a routing failure. Figure 4-6 shows the relationship between our example network. [18]. named ISPCust7. a brief discussion of CPP appears in Section 4. If support for RIP is not an essential requirement. [21]. MD5 authentication. Migration procedures are beyond the scope of this document. For more information on BGP and BGP Security. 4. In this particular case.1c 123 . Version 1. while the service provider gateway router. our example network constitutes a single autonomous system. OSPF offers better convergence times. Central(config)# no router rip Central(config)# After disabling RIP.3. but they do not show how to configure BGP routers in a large network. and the service provider that serves as its connection to the Internet. The example below shows how to turn off RIP. The examples in this sub-section show how to configure the indicated BGP security mechanisms. Both protocols are supported by virtually every routing vendor. Control Plane Policing (CPP) can also help protect BGP operations.4. [23]. and using OSPF reduces the likelihood of accidentally sending out routing update packets on an unintended interface. AS number 26625. However. an important step to remember is to remove RIP after OSPF has been enabled. This sub-section presents four security mechanisms that you should when using BGP-4: the Generalized Time-to-Live (TTL) Security Mechanism. Remember to turn off RIP on all the routers after migration. and is outside the scope of this guide.Advanced Security Services Migrating from RIP to OSPF: Security issues and concerns Although RIP has withstood the test of time and proven itself to be a reliable routing protocol.

0(27)S and 12.2. documented in RFC 3682 [32] and introduced in Cisco IOS 12. limiting the hop count to the actual number of hops between eBGP peers will help prevent attacks initiated on any network that does not lie between the peers. If the received IP packet contains a TTL value greater than or equal to the expected TTL value (i.2.2.3(7)T. This mechanism uses the TTL value in a received packet and compares it to an administrator defined hop count. North# trace ip 14.250 124 Version 1. Otherwise. otherwise the BGP session will terminate. use the trace command to verify that the BGP peers are indeed separated by the expected number of hops. [17]. the packet is silently discarded.e.1c . utilizes the Time-to-Live (TTL) field of the IP header to protect exterior BGP (eBGP) peering sessions from remote attacks.2. then the packet is processed.0. For more information on BGP operations. Before enabling GTSM. and [31].0. Generalized TTL Security Mechanism (GTSM) The Generalized TTL Security Mechanism (GTSM).0. Also.250 14.Router Security Configuration Guide 14. 255 minus an administrator defined hop count).20 North ISP Network ISPCust7 AS 27701 Local Network Other autonomous systems of the Internet AS 26625 Figure 4-6: BGP Neighbors and their Autonomous System Numbers This sub-section shows how to set up BGP authentication and other security measures. [21]. The following example shows how to implement the GTSM between routers North and ISPCust7 in Figure 4-6.0. consult [15]. it does not address operational issues that arise when deploying BGP infrastructure in a large network. Since remote attacks originate multiple router hops away from an intended target. Additional information on GTSM may be found in [33].250 Type escape sequence to abort Tracing the route to 14. coordinate the implementation of GTSM between eBGP peers so that the command sequence is issued on both peers during a time interval less than the BGP hold time.

e. spoofing.2. End with CNTL/Z.1. ISPCust7# config t Enter configuration commands. North (config)# router bgp 26625 North(config-router)# neighbor 14.20 0 msec * 0 msec ISPCust7# config t Enter configuration commands.0. one per line.2.250 ttl-security hops 1 North(config-router)# end North# ISPCust7# trace ip 14.20 1 14.20 ttl-security hops 1 ISPCust7(config-router)# end ISPCust7# BGP and MD5 Authentication BGP peers can be protected from DoS.0. End with CNTL/Z. North# config t Enter configuration commands. one per line. This security mechanism operates between two BGP peers and requires the configuration of a shared key on each of these peers.0. The shared key is used to create and verify the MD5 signature (i.1c 125 . one per line. ISPCust7(config)# router bgp 27701 Version 1.2. ISPCust7(config)# router bgp 27701 ISPCust7(config-router)# neighbor 14. End with CNTL/Z. “Protection of BGP Sessions via the TCP MD5 Signature Option” [34].2.20 Type escape sequence to abort Tracing the route to 14.0. The commands below would be performed on North by the local administrator. It shows how the network administrators of North in AS 26625 and the network administrators of ISPCust7 in AS 27701 would use the shared key “r0utes4All” when configuring MD5 authentication for their peering session. North(config)# router bgp 26625 North(config-router)# neighbor 14. BGP shared keys should follow the password guidance specified in section 4. one per line.Advanced Security Services 1 14. and man-in-the-middle attacks by implementing RFC 2385.0.250 remote-as 27701 North(config-router)# neighbor 14.2.2. The shared key takes the form of a password configured on each peer router.250 0 msec * 0 msec North# config t Enter configuration commands. End with CNTL/Z.5 and in the relevant IOS documentation The following two-part example pertains to Figure 4-6.2. one-way hash) for each segment transmitted and received during the BGP session respectively.250 password r0utes4All North(config-router)# end North# The commands below would have to be performed by the network administrators of the ISP router to which North has a BGP peering connection.

2. Cisco. AS ingress and egress prefix lists should filter martian address space (i. 5.e. There are several general rules that you should consider when developing prefix lists and filters for your BGP routers: 1. ISP egress prefix lists should prevent ISP core network prefixes from being advertised to any other AS. (Prefix filtering may be used to prevent an ISP from transiting traffic through a multihomed customer to another ISP.1c . Use prefix list filtering to enforce routing policy by disregarding advertisements and withdrawals of specific prefixes. Team Cymru.Router Security Configuration Guide ISPCust7(config-router)# neighbor 14.20 remote-as 26625 ISPCust7(config-router)# neighbor 14.20 password r0utes4All ISPCust7(config-router)# end ISPCust7# Prefix List Filtering Prefix list filtering is a common technique used to prevent damage such as DoS. ISP ingress prefix lists applied to advertisements from peer ISPs should prefixes that are more specific than those assigned by Internet registries (see [39] – [41]).0. including both RFC 1918 special use addresses [44] and bogon addresses [38] (i. 126 Version 1. 6.2. and prefix hijacking caused by malicious prefix advertisements. ISP ingress prefix lists applied to advertisements from a customer should permit only those prefixes that have been assigned or allocated to the customer. Although the use of prefix list filters is not restricted to BGP. 2.) Prefix list filters can be used to filter both inbound and outbound route updates on a per-peer basis. 4. traffic redirection. examples appear in [18] and [42]. AS ingress prefix lists should prevent internal prefixes from being advertised to the AS by some other AS. prefix list filtering is an important technique employed by AS network administrators to filter BGP route advertisements. address space that IANA has not yet allocated).e. address space that should never be propagated through the Internet). 3. and others maintain bogon address lists. At a minimum. ISP ingress prefix lists should deny advertised prefixes more specific than /24. These filters should only permit advertisement of routes to networks that have been allocated or assigned to the ISP and its customers. ISP egress prefix lists should filter what is advertised to other ISP peers.0.

from being advertised by router North and accepted by router ISPCust7. add/withdraw routes) to propagate through the network.0/12. 10. Version 1. These transitions cause excessive BGP route update messages (i.0/16 North(config)# ip prefix-list NO-MARTIANS seq 100000 permit 0. no other prefixes will be affected by the implicit deny of the prefix list filter mechanism.0.16. one per line.0. End with CNTL/Z. refer to [35] and [42].20 prefix-list NO-MARTIANS in ISPCust7(config-router)# end ISPCust7# Route Flap Damping Route flap damping is a method that may be used to provide router CPU and network stability while BGP routes are converging. ISPCust7(config)# ip prefix-list NO-MARTIANS seq 70 deny 172.0/12 ISPCust7(config)# ip prefix-list NO-MARTIANS seq 80 deny 192.0.1c 127 .0/8 North(config)# ip prefix-list NO-MARTIANS seq 70 deny 172. Damping controls the effect of route flapping which occurs when a route constantly transitions from an up-to-down or down-to-up state.168.168. ISPs and other backbone providers may configure BGP damping to mitigate route flapping. For more information on prefix filtering.e.2.0. North(config)# ip prefix-list NO-MARTIANS seq 60 deny It shows how to configure and apply prefix list filters to prevent a subset of martian addresses. ISPCust7(config)# ip prefix-list NO-MARTIANS seq 60 deny 10.e.250 prefix-list NO-MARTIANS out North(config-router)# end North# ISPCust7# config t Enter configuration commands.0/12 North(config)# ip prefix-list NO-MARTIANS seq 80 deny 192. notably private addresses (i.16. End with CNTL/Z.0. The explicit permit all prefix list entry with sequence number 100000 is used so that after the deny filters are applied.0.0/0 le 32 ISPCust7(config)# router bgp 27701 ISPCust7(config-router)# neighbor 14. and 192.0/16).0/16 ISPCust7(config)# ip prefix-list NO-MARTIANS seq 100000 permit 0.0. North# config t Enter configuration commands. The sufficiently high sequence number 100000 is chosen so that many new entries can be added to the prefix list sequentially prior to the explicit permit all entry. one per line.Advanced Security Services The following example pertains to Figure 4-6.0.0/0 le 32 North(config)# router bgp 26625 North(config-router)# neighbor 14.0.

depending on the network 128 Version 1. use one of the two commands shown below. default is 750. The fine granuality access lists provide can impose significant administrative and performance burdens. [37]. For example. Opinions differ on the usefulness of route flap damping and the correct values for damping parameters. default is 15 minutes. If you choose to use route flap damping.20 remote-as 26625 bgp dampening end The selection of BGP damping parameters for use in the Internet is not trivial and has been the topic of debate for several years. bgp dampening [half-life] [reuse] [suppress-limit] [maxsuppress-time] • half-life . Consult the following references before implementing route flap damping: [23]. Using Black-Hole Routing Many administrators configure their routers to filter connections and drop packets using basic and extended access lists.g. • reuse .1c . To display the dampened routes with the corresponding suppression time remaining. • max-suppress-time . • suppress-limit .4. and monitor your network carefully to determine the appropriate parameters for your network environment.2 and earlier ! 12.3 or North# show ip bgp dampening dampened-paths 4. The list of dampened paths is useful in determining which remote networks are having instability problems.range is 1-20000. 60 minutes). begin by using the defaults. The following example shows the network administrator of ISPCust7 in AS 27701 enabling route flap damping with default parameter values.2. [36].0. North# show ip bgp dampened-paths ! 12.range is 1-20000.range is 1-45 minutes. Operational experience has shown that vendor defaults may be too aggressive. Access lists provide the administrator with a high degree of precision in selectively permitting and denying traffic. default is 4 times the value of the half-life parameter (e. ISPCust7(config)# router ISPCust7(config-router)# ISPCust7(config-router)# ISPCust7(config-router)# ISPCust7# bgp 27701 neighbor 14. default is 2000. access lists would allow an administrator to block only Telnet (TCP port 23) traffic from exiting their network. Cisco default parameters are shown below.6.Router Security Configuration Guide The syntax for the bgp dampening command permits several optional parameters.range is 1-225 minutes.

It is possible to use a selective BGP or OSPF neighbor router to distribute Version 1. For every packet that was filtered.0 null0 Central(config)# exit Central# Enter configuration commands. Because the null0 interface is a packet sink. it can be used only to impose a ban on all traffic sharing a specific destination address or network. An alternative to access lists for traffic control is a technique known as black hole routing. the following configuration would work: Central# config t Central(config)# interface null0 Central(config-if)# no ip unreachables Central(config-if)# exit Central(config)# ip route 10. Null routing sacrifices the fine selectivity of access lists. in particular. This makes it well-suited to mitigating attack situations where ‘bad’ traffic into your network is all directed to one or a small number of address ranges. It is important to turn off the generation of ICMP unreachable messages on the null0 interface. Configuring Null Routing The simple way to configure null routing is to set up a null interface and create a static route that directs the undesirable packets to it. It is important to note that null routing can only discard traffic based on its addresses (usually only the destination). the offending packets will be dropped silently. On a Cisco router.Advanced Security Services architecture. one per line. Backbone routers. router configuration. If an address or network is null routed. If an administrator was utilizing null routing to block a denial of service attack. There is no simple way to specify which protocols or types of traffic may or may not pass.0/8 network. packets sent there will never reach their intended destination. and traffic load. ALL traffic sent to it will immediately be discarded. More sophisticated filtering with null routing is possible with more advanced techniques. this would cause the router trying to block the attack to ultimately flood its own upstream with ICMP unreachable messages.0.0. using ip route statements as shown above. are often too heavily utilized to permit heavy use of access lists. This can compound the damage of the initial attack. or null routing.0. To null route additional IP addresses in the future. the router would send a message back to the host originating the attack. It is also well-suited for discarding data directed to unassigned or reserved addresses. End with CNTL/Z.0. to block packets with a destination address in the reserved range of 10. For example. When you disable ICMP unreachable messages.0 255.0. Because this type of filtering is done as part of normal routing. it imposes little or no performance burden on normal packet flow. the default behavior when a packet cannot be delivered to its intended destination is to send the source address an ICMP unreachable message.1c 129 .0. you would simply add additional static routes.

However. you must explicitly apply it to each interface where you want verification to be done. It finds interface Eth 0/0.4. it fails unicast reverse-path verification. Eth0/1. Unicast reversepath verification is not enabled by default. in the routing table. When this feature is enabled on an interface.20. Figure 4-7 shows two packets arriving at the router Central on its ‘inside’ interface. the router looks up its source address. 14. Null routing can also be combined with filtering to support traceback of some types of DoS attacks. in the route table.12. so the router forwards packet 1 normally out interface Eth 0/0.B.Router Security Configuration Guide null routes throughout a network. Used correctly. then the packet is considered ‘good’. and the router discards the packet.D.1. it is good security practice to reject a packet with a spoofed source address. bad packets are discarded. it might have been generated by a piece of malicious software secretly installed somewhere on LAN 2. It is also possible to set up automatically triggered null routing in an entire network. 7.B. the configuration of such advanced null routing is beyond the scope of this guide. Good packets are forwarded normally. the router uses its routing tables to decide whether to accept or drop individual packets arriving on the interface.C. 130 Version 1.C. Because the packet has arrived on the wrong interface. For packet 2.D was received is the one that the router would use to send a packet to A.0 and later support a routing-based filtering feature called IP unicast reverse-path forward (Unicast RPF) verification. for a detailed look at the topic.2. This is a match. It finds the interface Eth 0/1. The first packet bears a proper source address.3.10. For packet 1. Additional practices for null routing are described in [15]. As noted in Section 4. otherwise it is ‘bad’. unicast RPF verification prevents most forms of IP address spoofing. Unicast reverse-path verification uses the routing table to decide whether a packet with a particular source address is valid: if the interface on which the packet with address A. the router looks up its source address. consult [45]. Setting up such a capability is beyond the scope of this guide. Unicast Reverse-Path Forwarding Verification Most Cisco routers running IOS 12. The second packet bears a spoofed source address. it is from a host behind the South router. and in situations where it applies.1c .7. 4. Unicast reverse-path verification supports rejecting such packets. How Does Unicast Reverse-Path Verification Work? All routers maintain a routing table that lets them decide how to forward packets. which is not the interface on which packet 2 has arrived. as described in [24]. and in some cases it can offer significant advantages over using access lists for that purpose. which is the interface on which packet 1 has arrived.2.

0/24 Pa c ke t 2 Interface Eth0/0 Trash Interface Eth0/1 Packet 1 src=14.Advanced Security Services -14.250 14.250 Central Eth0/1 14.0/16 Eth0/0 14.1.0/0 -14. • Router does not support CEF – according to the Cisco documentation.6. but only when the network architecture permits it to be used.1. unicast reverse-path verification depends on Cisco Express Forwarding. Avoid unicast RPF verification if any of the following conditions apply. When to Avoid Unicast Reverse-Path Verification This facility can be very useful for rejecting packets with improper IP source addresses.20 Packet 1 Destination Gateway Interface 14.12. Version 14.1.0. and a different one for receiving). also require more maintenance. use access lists instead.20 dest=7.15. If your router does not or cannot support CEF.1.1. while more broadly applicable. • Router uses asymmetric routes – if any of the interfaces on the router participate in asymmetric routes (one interface for sending.12.1.0/24 14.1.64 14.2 dest=10. then you cannot use unicast RPF.0/24 14.0/16 0.1c 131 .9. it automatically adjusts to most changes in network structure.0.9 Packet 2 src=7. then simple unicast RPF verification may not be used. Cisco has stated that future IOS versions will perform unicast RPF correctly in these cases [11]. Eth 0/1 Eth 0/1 Eth 0/0 Eth 0/0 Eth 0/0 Routing Table Figure 4-7: IP Unicast RFP Verification Because unicast RPF verification uses the route table. Access lists. It will incorrectly reject packets arriving on the receive leg of the asymmetric route.

to use unicast RPF. consult [10]. If the access list denies the packet.3 and later). Under 12. When a packet fails reverse-path verification.1c .3. only the lines about drops may appear.255. IP verify source reachable-via RX. Thus. Central# show ip interface eth0/0 Ethernet0/0 is up.255. In particular.15.250/16 Broadcast address is 255. use show ip interface interface-name.Router Security Configuration Guide Unicast RPF verification is best suited for routers that act as part of the security boundary between two networks (e.1.2 and earlier. Also. it can provide better performance than an access list for ingress and egress address filtering. The transcript below shows how to enable verification on the router Central. Central# config t Central(config)# ip cef Central(config)# interface eth 0/0 Central(config-if)# ip verify unicast reverse-path Central(config-if)# end Central# Cisco routers equipped with Versatile Interface Processors (VIPs) may require you to enable CEF with the command ip cef distributed instead of the simple version shown above. Consult [10] for details about CEF requirements.255 . The example below shows how the output will look when unicast RPF is enabled in IOS 12. Used properly. enter interface configuration mode and use the command no ip verify unicast reverse-path (12. If the access list permits the packet. . or to view statistics about dropped packets. a filtering router between a LAN and the Internet). and then enable verification on the desired interfaces.1 and later include significant enhancements to unicast RPF. line protocol is up Internet address is 14.g. Configuring Unicast Reverse-Path Verification Unicast RPF verification depends on a particular routing mode called Cisco Express Forwarding (CEF). Unicast Reverse-Path Verification and Access Lists Cisco IOS version 12. To check whether unicast RPF is enabled on a particular interface.0 through 12. first enable CEF. For more details on how and where to apply unicast RPF verification. the access list allows you to create exceptions to unicast RPF’s usual functioning. Note that you must not turn off CEF while unicast RPF is enabled. allow default 0 verification drops 0 suppressed verification drops Central# To disable unicast RPF. 132 Version 1. then it is dropped. then it is forwarded. Therefore.2) or no ip verify unicast (12. then the access lists are applied. access lists may be applied to RPF.

Comprehensive and practical guide to OSPF use. Advanced Cisco Router Configuration. TCP/IP Illustrated.. 1998. Coriolis Group.Design and Implementation Guide. W. Includes discussion of design issues. Includes a good section on troubleshooting. great technical background for any network analyst. including routing and routing protocols. OSPF Network Design Solutions. A very practical and pragmatic guide to setting up routing protocols.. IP Routing Protocols. [6] Stevens. 1994. 1999.1c 133 . [4] Moy.. This book offers a good overview of IP routing and related topics.M. with lots of practical advice. W. 1998. Addison-Wesley. implementation. J.8. Editor. T. Cisco Routers for IP Routing: Little Black Book. consult [11] and [15].1) then a log message is generated. [2] Parkhurst. 1999. Addison-Wesley.. and also explains how to configure Cisco routers for OSPF in a variety of situations. Version 1. Prentice Hall. [8] Rudenko. 4.. Cisco Press. U. A great reference book for a variety of Cisco configuration topics. OSPF – Anatomy of an Internet Routing Protocol.. Cisco IOS Essentials.R. The most comprehensive and readable guide to the TCP/IP protocol suite. For more information about advanced unicast RPF features. Volume 1. too.R. and deployment. Detailed analysis of OSPF and MOSPF. McGraw-Hill..4. security. I.Advanced Security Services if the access list rule that denies a packet includes the log qualifier (see Section 4.T. Cisco Press. Cisco Router OSPF . J.3. [5] Thomas. 1999. An excellent introduction to basic Cisco IOS tasks. [3] Black. Portions of this book that are particularly relevant to Routing Protocols are Chapters 2 and 7. 1998. [7] Chappell. A very good survey of routing protocols and the technologies behind them. McGraw-Hill. References [1] Albritton. Laura. 2000.

and an in-depth treatment of Unicast software/ios121/121newft/121t/121t2/rpf_plus. The IETF specification for Proxy ARP. Cisco Router Troubleshooting Handbook.. The IETF specification for ARP. available at: http://www. J. [16] “Routing Basics”. Cisco Systems. and Smith. available at: http://www.htm As a general overview of routing concepts and terminology. all with fully worked-out examples. available under http://www. 2000. 2000. Cisco Internetworking Technology Overview.1(2)T Release Notes. 2002. Cisco IOS 12.htm Documentation for new Unicast RPF features that are being integrated into IOS 12. [13] RFC 826. D. 1982. 134 Version 1. This pragmatic volume offers good advice for diagnosing and correcting problems with routing and routing protocols. this document gives a broad. 1st Edition. RFC 1027. “RIP and OSPF Redistribution”. [14] Rybaczyk.Router Security Configuration Guide [9] Cisco Systems. [12] [10] “Configuring Unicast Reverse Path Forwarding”.0 Security Configuration Guide.pdf Basic IOS documentation on unicast reverse-path forwarding verification. available at: http://www..1c . Cisco Systems. C-M. performance-oriented view of IP routing. Cisco Press. April 2002. “An Ethernet Address Resolution Protocol of Converting Network Protocol Addresses to 48-bit Ethernet Address for Transmission on Ethernet Hardware”. and This detailed Cisco guide for Internet Service Providers includes extensive discussion of routing protocols (especially BGP)..1 releases. “Using ARP to Implement Transparent Subnet Gateways”. [15] Greene. P. 1987. Cisco IOS 12. [11] “Unicast Reverse Path Forwarding Enhancements”. Cisco Systems. Cisco Internetworking Case software/ios120/12cgcr/secur_c/scprt6/scrpf. M&T ito_doc/routing. includes a good explanation of the concepts. Cisco ISP Essentials. B.. Cisco Systems. 2000.

Cisco Systems. troubleshooting and verification guidance. [23] Panigl.1”.org/Tracking/ This terse and technical note describes a technique for using null routing. RIPE.. [22] Cisco IOS IP Configuration Guide. BGP4 . Provides a basic understanding of routers and routing protocols through a thorough inspection of IP interior gateway routing protocols.. Cisco Press. concise resource for Cisco IOS software BGP-4 commands. Version 1.Inter-Domain Routing in the Internet.R.Volume 1. [20] “Enhanced IGRP”. BGP. 2001. The bookemphasizes techniques for designing efficient networks. A clear. and access control lists to trace back the sources of some DoS attacks. Schmitz. RIPE-229. A white paper that describes the features and operation of EIGRP. 1998. available at: http://www. [24] Morrow.html This note describes rationale for BGP route flap dampening.secsup. Version 2. [21] Parkhurst. [19] Doyle. Routing TCP/IP . Cisco Press. October 2005. C. J. Smith. 1999. R. Cisco Press. [18] Thomas. available at: http://www. The command guide provides very good configuration.. Cisco Internetworking Technology Overview. and Vistoli..0.1c 135 . 2002. an excellent reference for network engineers responsible for enterprise design. W. This volume of the Cisco IOS documentation offers extensive information on configuring all of the routing protocols discussed in this section.ripe.Advanced Security Services [17] Stewart.html This short but highly prescriptive document gives a detailed example of a locked-down configuration for a backbone or border router using BGP-4. October 2001. Addison-Wesley. and includes links to examples for Cisco IOS. 2002. J. Cisco BGP-4 Command and Configuration Provides a good overview of BGP and practical advice on using it. “Secure BGP Template Version 4.2. “RIPE Routing Working Group Recommendations for Coordinated Route-flap Damping Parameters”. “BlackHole Route Server and Tracking Traffic on an IP Network”.W. available at: IOS 12.

F. 2002. August 1998. [31] Zhang. IS-IS Network Design Solutions. “The Generalized TTL Security Mechanism (GTSM). Cisco Technologies Documentation. [27] Martey. Cisco Systems. Internet Routing Architectures .cisco. and Meyer. The definitive IS-IS reference and design guide. 2001.html This page offers access to a large number of technical documents and examples about IOS-supported routing protocols. and Bartell.pdf Provides a good overview of IS-IS authentication with an emphasis on plaintext authentication. Cisco Systems. Troubleshooting IP Routing Protocols. 12... M.. Cisco Systems. [32] Gill. A thorough presentation of inter-domain routing concepts. BGP Design and Implementation. A. Good reference describing the IOS command set for HMAC-MD5 authentication.. [26] “IS-IS HMAC-MD5 Authentication and Enhanced Clear Text Authentication” IOS Release Notes for 12. Cisco Press.Router Security Configuration Guide [25] “IP Routing Protocol Groups”.1c .” RFC 3682. Cisco IOS Release Notes for 12. The IETF specification for GTSM. A reference for IOS configuration of the GTSM. February 2004. Cisco Press. available under: http://www. A comprehensive hands-on guide for resolving IP routing problems. Cisco Systems. “Protection of BGP Sessions via the TCP MD5 Signature Option. Heasley. A. including guidance on multiprotocol extensions. Cisco Technology White Paper. 2001.Second Edition. 2002. organized by protocol. [30] Halabi.0(22)S. 2003.2(25)S. and scenarios.. S. 136 Version 1.0(21)ST and 12.3(7)T. [33] “BGP Support for TTL Security Check”. and 12. Cisco Press. Cisco Press. [28] “Configuring IS-IS Authentication”.0(27)S. A discussion of advanced BGP topics applicable to both enterprises and ISPs. R. available at: tsd_technology_support_protocol_home. containing command syntax and a detailed configuration procedure. 2004. [29] Shamin.” RFC 2385. [34] Heffernan.

ripe. available at: Asia Pacific Network Information Center.” Proceedings of the RIPE50 Conference. Griffin. ARIN. for use in prefix list An excellent resource about bogons and bogon ripe-ncc-managed-address-space. [38] “The Team Cymru Bogon Reference Page”. available at: P.html A list of minimum practical allocations by RIPE. Ingress-Prefix-Filter-Templates/Prefix-List-Filtering-Mar-03.. and Mao. Also includes some explanatory notes. Team Cymru. Cisco course notes.pdf A good overview of prefix filtering with examples and references. available at: http://www. available at: ftp://ftp-eng.Advanced Security Services The IETF specification for MD5 protection of BGP sessions. ISP Security Boot Camp. 2005. [41] “Allocation sizes within APNIC Address Ranges”. “Flap Damping .1c 137 .net/ripe/meetings/ripe-50/ presentations/ripe50-plenary-wed-flap-damping. This page also provides a well-maintained list of bogon address ranges in a variety of formats. 2005. [37] Smith.html A list of minimum practical allocations by APNIC. available at: A conference presentation that questions the value of route flap damping. available at: http://www. 2002. L. 2005.Where to now?.html#cidr An overview of address space allocated by ARIN. available at: http://www. RIPE 349. with links to detailed lists and historical data. “Address Space Managed by the RIPE NCC”.arin. B.pdf A presentation of flap damping issues and a call for a response to deal with those issues. “Route Flap Damping: Harmful?” NANOG presentation. [39] “ARIN 2005 Statistics”. [36] Bush. 2005. Cisco Systems. 2005. for use in prefix list filtering. [40] Vegoda.apnic. [35] Greene. Version 1.

B. available at: http://www.pdf A recent and detailed paper about threats to routing protocols and means to mitigate those threats. Maintained as part of the bogon project [38]. Cisco Systems. [44] “SAFE: Best Practices for Securing Routing Protocols”. RFC 2004. D. 138 Version 1. Y. “Ingress Prefix Filter Templates”. Cisco white paper. Moskowitz. to be used for filtering and Rekhter. July 2005. [45] “Remotely Triggered Black Hole Filtering – Destination Based and Source Based”. available under: ftp://ftp-eng. available at: http://www.pdf An in-depth description of how to set up centrally-triggered null routing using your internal BGP mesh. February 1996. version 1. This RFC describes the reserved IP address allocations for private intranets. 2005.4.1c . B. [43] docs/ Ingress-Prefix-Filter-Templates/ IOS command templates for ingress prefix filter prodlit/sfblp_wp. Cisco Systems. Cisco white paper. Cisco Systems. “Address Allocation for Private Internets”.Router Security Configuration Guide [42] Greene.

Concepts and Mechanisms Routers are a critical part of network operations and network security. • Software Maintenance – Keeping up with new major software releases is important. Careful management and diligent audit of router operations can reduce network downtime.5. good logs can help you find configuration errors. improve security. NTP. Cisco routers support the standard time synchronization protocol. 12. • Time – Accurate time is important for good audit and management. describes the logging configuration commands. Complete details on the commands and features discussed may be found in the Cisco IOS documentation. SNMP must be disabled or carefully configured for good security.5. Installing new Cisco IOS software in a router is not especially difficult. • Network Monitoring – Cisco routers support basic facilities for Remote Network Monitoring (RMON). • Logging – Cisco routers support both on-board and remote logs. This section describes the logging. understand past intrusions. 4. The RMON features depend on SNMP. troubleshoot service disruptions. and update facilities offered in Cisco IOS 11.2. this section explains the different logging facilities.0.0. and react to probes and scans of your network. Version 1. because new releases include fixes for security vulnerabilities. and must also be disabled or carefully configured. and aid in the analysis of suspected security breaches. The sub-sections below describe recommended configurations for good security.5. management.3. Configuring Logging and Time Services Logging is a critical part of router security. and presents some configuration examples. • Debugging and Diagnostics – Troubleshooting router problems requires proficiency with Cisco’s diagnostic commands and debugging features.1c 139 . and later. Cisco routers and Cisco IOS are designed to support centralized audit and management.Advanced Security Services 4. • Network Management – The standard protocol for distributed management of network components is the Simple Network Management Protocol (SNMP). monitoring. especially the Cisco IOS Configuration Fundamentals Command Reference documents for IOS 12.1. Cisco routers have the ability to log a great deal of their status. Audit and Management 4.

but does not offer enough long-term protection for the logs. Some motivations for keeping router logs are listed below. This form of logging is useful. but are otherwise of little value unless some other device or piece of software preserves the output. Buffered logging – Cisco routers can store log messages in a memory buffer. This form of logging is not persistent. and many more kinds of events.1. that can be preserved over time. Syslog logging – Cisco routers can send their log messages to a Unix-style syslog logging service. messages printed to the console are not stored by the router. or any combination. changing a password. Four such events are: changing EXEC privilege level. which is used on the Internet and on all major US DOD networks to distribute accurate time. Messages can be sent to all five. Configuration guidance for NTP appears at the end of this sub-section. 2. The buffered data is available only from a router exec or enabled exec session.2) There are some events that can be important to security but which Cisco routers cannot log. Terminal Line logging – Any enabled exec session. changes in network and interface status. This form of 140 Version 1. as discussed below. and it is cleared when the router boots.2). Overview and Motivations for Logging Cisco routers can log system errors. The most valuable forms of logging are forms that are persistent. 3. 4. on any line.3) • Recording changes in interface and network status • Recording router cryptographic security violations (see Section 5. Turning on line logging is useful only for the operator using that line. A syslog service simply accepts messages. Console logging – Log messages are sent to the console line (see Section 4. 1. This form of logging is not persistent. changing the configuration via SNMP. and saving a new configuration to the NVRAM.Router Security Configuration Guide Keeping the correct time on a router is also important for accurate logs. Console logging is handy for operators when they use the console. can be configured to receive log messages.1c . Log messages can be directed in five different ways. • Recording router configuration changes and reboots • Recording receipt of traffic that violates access lists (see Section 4. login failures. and stores them in files or prints them according to a simple configuration file. access list matches. Cisco routers fully support the standard Network Time Protocol (NTP).

Cisco IOS messages are categorized by severity level.Advanced Security Services logging is the best available for Cisco routers. you must use the level name. as shown by the numeric field “-5-” in the message name. Note that. Table 4-3 – Cisco Log Message Severity Levels Level 0 1 2 3 4 5 6 7 Level Name emergencies alerts critical errors warnings notifications informational debugging Description Router becoming unusable Immediate action needed Critical condition Error condition Warning condition Normal but important event Information message Debug message Example IOS could not load Temperature too high Unable to allocate memory Invalid memory size Crypto operation failed Interface changed state.9. It has a severity level of 5. Cisco routers can generate Simple Network Management Protocol (SNMP) trap messages. the message below appears in the log when a user changes the running configuration. because it can provide protected long-term storage for logs. 5. The severity levels are described in the table below. Mar 31 9:00:16 EST: %SYS-5-CONFIG_I: Configured from console by vty0 (14. when you are using logging levels in commands in IOS 11.3 and earlier. up to down. the more critical the message is. SNMP trap logging – For some kinds of events.6) Message text Message name and severity level Time that message was generated Figure 4-8: Format of a Cisco IOS Log Message Version 1. This facility allows routers to be monitored as part of an overall SNMP-based network management infrastructure.2. in IOS 12.1c 141 . The lower the severity level number.0 you may use the name or the number. or down to up Packet denied by an access list on an interface Appears only when debugging is enabled For example.

Central(config)# logging on End with CNTL/Z Setting up Console and Buffered Logging To turn on console logging. one per line. below.1c . consult the “Troubleshooting Commands” section of the IOS Configuration Fundamentals Command Reference.5. one per line. buffered logging. Cisco routers have the ability to timestamp their messages. as shown below.4 presents RMON configuration issues. enable SNMP trap logging also.3. Central# config t Enter configuration commands. but be aware that this can place a burden on the router and should be used sparingly. but it must be turned on explicitly. Use the command logging console info to see all non-debug messages including access list log messages. your log buffer size should be about 16 Kbytes. Central(config)# ! set console logging to level 5 (notify) Central(config)# logging console notification Central(config)# exit This example sets the console message level to 5. and view the buffered log.) The descriptions below recommend logging configuration settings.5. then you can set the log size to 32 or 64 Kbytes. set up syslog logging. This example sets the logging level for the console to level 5. As a rule of thumb. sub-section 4. notifications. RMON is a monitoring facility based on SNMP. and consider use of console logging. Central# config t Enter configuration commands.Router Security Configuration Guide For best security. Use logging console debug to see ALL messages on the console. For buffered and other forms of persistent logs. recording the time and date of the logged message is very important. In a network where SNMP management is already deployed. but access list log messages will not. which means that important messages will appear on the console. End with CNTL/Z Central(config)# ! Set a 16K log buffer at information level Central(config)# logging buffered 16000 information Central(config)# ! turn on time/date stamps in log messages Central(config)# service timestamp log date msec local show-timezo Central(config)# exit Central# show logging 142 Version 1. First. the logging level at the console should be set to display lots of messages only when the console is in use or its output is being displayed or captured. use the commands shown below. if your router has more than 16 Mbytes of RAM. If you are not using the console. turn on logging services. set the console logging level to 2 using the configuration command logging console critical. enable time stamps. In general. The example below shows how to turn on buffered logging. (SNMP is discussed in sub-section 4. for more information about Cisco logging command and facilities.

2.Advanced Security Services Syslog logging: enabled (0 messages dropped.1c 143 .1 flushes. while using a particular line. The example below shows how to set up terminal line monitoring for informational severity (level 6) on a telnet session virtual terminal line.6 Central# config t Enter configuration commands. Central# show users Line User Host(s) Idle Location *130 vty 0 bob idle 00:00:00 14.9.2. one per line. It offers the network administrator the ability to send log messages from all of the routers (and other Cisco equipment) on a network to a central host for examination and storage. set the severity level for terminal line monitor log messages. . There are two parts to setting up terminal monitor logging. 302 message lines logged Log Buffer (16000 bytes): Mar 28 11:31:22 EST: %SYS-5-CONFIG_I: Configured from console by vty0 (14. level 5 Central(config-if)# shutdown Central(config-if)# Mar 28 15:55:29 EST: %LINK-5-CHANGED: Interface Ethernet0/1.2. End with CNTL/Z. For more information consult the tools list in Section 9.3. Version 1. changed state to administratively down Setting up Syslog Logging Syslog logging is the most useful form of logging offered by Cisco routers.6) . Central(config)# interface eth 0/1 Central(config-if)# ! shutdown will log a message. 332 message lines logged Logging to 14. Central(config)# ! set monitor logging level to level 6 Central(config)# logging monitor information Central(config)# exit Central# ! make this session receive log messages Central# terminal monitor Central# config t Enter configuration commands. Central# Setting up Terminal Line Logging Any terminal or virtual terminal line can act as a log monitor. this needs to be done once per session.9. and several free and commercial syslog servers are available for Windows NT/2000/XP. one per line.0 overruns) Console logging: level critical. End with CNTL/Z. Second.6.9. this needs to be done only once. All Unix and Linux operating system configurations include syslog servers. 1 messages logged Trap logging: level debugging. 0 messages logged Buffer logging: level informational. declare it to be a monitor. First.

Set the severity level limit for messages sent to syslog using the logging trap command.1c . There are several dozen valid syslog facility names. Syslog servers also support the notion of severity levels. The diagram below shows a typical configuration with syslog in use.Router Security Configuration Guide Review of Syslog Concepts A syslog server is a network host that accepts messages and processes them.2. Any number of syslog hosts may be specified.15. The severity level for syslog messages is usually the same as that for buffered log messages.64 Eth 0/0 14.250 Eth 0/1 LAN 2 14.9.2. 144 Version Central log messages log messages 14. The syslog facility is simply the name you’ll use to configure storage of your messages on the syslog server.10. The source interface is the network connection from which the syslog messages will be sent. the levels have the same meanings as the Cisco severity levels listed in Table 4-3 above. and the source interface for the messages. the syslog facility . use the loopback interface if you have defined one. or an IP address.9. The destination host may be specified with host name.250 14. A syslog client is a host that generates messages. a DNS name. but the ones used for routers are typically local0 through local7. the log severity level.2. for more information consult any Unix syslog.1. otherwise use the network interface closest to the syslog server.conf(4) manual page or other syslog documentation on the server host.6 South Eth 0/1 Syslog server host Log message storage LAN 3 Figure 4-9: A Small Syslog Configuration There are four things that you must set for syslog logging: the destination host or hosts. LAN 1 Eth 0/0 14. but typically only one or two are needed (see below).

At least one of the two syslog servers’ logs should be backed up to permanent storage (CD-R or tape). The syslog configuration lines below illustrate this.log local6. For example. set up access control lists to reject syslog traffic from the outside network. Syslog uses UDP port 514. The example below shows the syslog configuration line for saving Central’s messages into a file.9.2. 11 flushes.9.1c 145 . separate the devices into 2-5 populations with similar duties. such as a border router on the Internet. it is best to designate two independent syslog servers. local6 for border routers. Configuration file syntax for syslog servers is uniform for all Unix and Linux syslog servers. local5 for interior routers. consult Section 4.2.255 eq syslog access-list 120 deny udp any 14.9.0.debug /var/log/routers.10.log Additional Issues for Syslog Logging For a router whose security is critical.6 Central(config)# logging facility local6 Central(config)# logging source-interface loopback0 Central(config)# exit Central# show logging Syslog logging: enabled (0 messages dropped. using syslog facility local6 and the loopback interface. one per line. 28 message lines logged . Central# Central# config t Enter configuration commands. access-list 120 deny udp any 14. .2. End with CNTL/Z. # Save router messages to routers. Save all messages of critical (level 2) severity and above to a single special file.0. Central# It is important to configure the syslog server to store router messages in their own file.0.6. Version 1.255 eq syslog In a situation where a sizable set of routers and other devices are sending messages to the same syslog server.0 0. For more information on access lists.3. Use a separate syslog facility name for each population. and otherwise save messages for each facility into a separate file. shown in the figure above. rather than denying specific ports as shown here).conf. 35 messages logged Monitor logging: level debugging. 0 overruns) Console logging: level notifications. Central(config)# logging trap information Central(config)# logging 14.Advanced Security Services The example below shows how to configure the router Central. 31 messages logged Logging to 14.0 0.2. 35 messages logged Buffer logging: level informational. On a border router. An example access list entry for the router Central is shown below (note: it is usually better to set up your access lists to permit explicitly required ports and protocols and deny all else. to load informational severity and above (level 6) messages to the syslog server.0. and local4 for LAN switches and other network hardware. the configuration file is almost always /etc/syslog.

1. last. It may be necessary to add Ciscospecific format specifications to the SNMP tools.1 traps public Central(config)# snmp-server trap-source loopback0 Central(config)# snmp-server enable traps syslog Central(config)# exit Central# Many of the trap messages sent by a Cisco router will not appear as formatted error messages in commercial SNMP viewing tools. although some older Cisco models may lose time when turned off.log local4. End with CNTL/Z. The example below shows how to configure SNMP trap logging for a receiving host 14.crit /var/log/net-critical. It is possible to perform manual network time synchronization.debug /var/log/border-routers. set the SNMP source interface.2.crit /var/log/net-critical.log local5. and may be useful in monitoring the network status. trap messages about link status changes and other typical network hardware events should be interpretable by commercial SNMP tools.log # All other router and switch messages to their respective files local6. Time Services.9. it may be useful in a network that already has SNMP management deployed. and then keep the time synchronized while the router is in operational use.log local4.9. SNMP is described in more detail in the next sub-section. enable SNMP traps for syslog logging. set the trap logging level.debug /var/log/inner-routers. Network Time Synchronization and NTP Successful audit of a large network can depend on synchronization of the various logs and records maintained for the hosts on that network.crit /var/log/net-critical. Central# config t Enter configuration commands. and unreliable. While only a small subset of all log messages can be reported this way. There are four parts to setting up SNMP trap logging.log SNMP Trap Logging Cisco routers have the ability to report certain events as SNMP traps. second.debug /var/log/other-net-hw. third. error prone.Router Security Configuration Guide # Critical and higher messages to critical. It is very important to set the time on a router when it is first installed. and no router can keep accurate time by itself over weeks and months of operation. select an SNMP logging host.2. All Cisco routers have a clock that maintains the time and date. Cisco routers fully support automated network time synchronization based on the standard Network Time 146 Version 1. one per line.log local6. adjusting the time on each router and host on a network manually on a regular basis. However.1c .log local5. Manual time synchronization is tedious. First. Central(config)# logging trap information Central(config)# snmp-server host 14.

as shown in Figure 4-10: one or more stratum 1 servers get their time from an authoritative source. NTP should be used to synchronize all the devices and hosts on a network whenever it is available.Advanced Security Services Protocol (NTP). and so on. one per line.1c 147 . Central# config t Enter configuration commands. If the timezone is not correct. and explain how to configure it on Cisco IOS. set the timezone if necessary. The sub-sections below give some background information on NTP. especially for intrusion and forensic analysis. US east and west coasts) then you should set the router time zone on all your routers to universal time or GMT.747 UTC Tue Mar 28 2000 Time source is user configuration Central# config t Enter configuration commands. Examine the clock using the show clock detail command.495 EST Tue Mar 28 2000 Central# If you manage routers spread across several time zones (e. End with CNTL/Z. Setting the Time Manually To set the time. second. and it is used on most large operational networks. If the detail output reports a time source of NTP. Central(config)# clock timezone GMT 0 Central(config)# exit Central# Review of NTP Concepts NTP is the standard Internet protocol for time synchronization. Central(config)# clock timezone EST -5 Central(config)# interface eth 0/0 Central(config-if)# ntp disable Central(config-if)# end Central# clock set 17:27:30 28 March 2000 Central# show clock 17:27:34. Version 1. Otherwise. check the clock. then set the time zone using the clock timezone configuration command. see the descriptions of NTP below. set the time in privileged EXEC mode by using the clock set command. Central# show clock detail 22:26:21. and turn off NTP on each interface using ntp disable.g. one per line. then do not set the clock manually. Stratum 2 hosts get their time from stratum 1 servers. End with CNTL/Z. Because having accurate time can be important for security. follow these three steps: first. NTP is designed to make time synchronization automatic and efficient. like an atomic clock or GPS. and last set the time. Typical NTP deployment is hierarchical.

ntp.Router Security Configuration Guide Atomic Clock Stratum 1 Stratum 2 Stratum 3 Stratum 4 Figure 4-10: The NTP Hierarchy If an NTP client is configured with several NTP servers. This configuration will allow the router to block general NTP traffic at the boundary. For more information about NTP configuration. consult the “Performing Basic System Management” chapter of the Cisco IOS Configuration Fundamentals Configuration Guide. NTP authentication should also be used (see below). setting up peer associations is beyond the scope of this see [11]. such as the ntpq tool distributed with the open-source NTP package. the router should be configured as an NTP client to two or more reliable Internet time servers. a Cisco router may be used as the border router between the Internet and an internal. In some cases. and may serve as the NTP server to the hosts on the internal network. protected network which requires time synchronization from a time server on the Internet. Cisco routers (except the old 1000-series) are capable of acting at any stratum in the NTP hierarchy except stratum 1. Note that Cisco IOS implements version 3 of the NTP protocol. visit the main NTP site: http://www. If your network has one of these. NTP clients may also have peer associations. it will select among them automatically based on time accuracy and stratum level. 148 Version 1. Commercial stratum 1 radio receivers are available that use a broadcast time source (e. If at all possible. For more information about NTP and Internet NTP servers. In these cases. and can be monitored using NTP standard-compliant tools.g. then you can configure all your routers to get their time from it. GPS or a time signal from the US Naval Observatory) to offer NTP service. For more information on this topic. or to obtain the latest NTP server software and tools for a variety of operating systems. As shown in the figure.1c . directly or indirectly.

set the NTP source interface.1c 149 .Advanced Security Services Configuring Basic NTP Service To set up a Cisco router to participate in an NTP network. use the network interface on the same network as the designated server. or the one that is the fewest number of network hops distant from the servers.2.250 eq ntp host 14.170 EST Wed Mar 29 2000 Time source is NTP South# Access restrictions can be imposed on NTP in two ways: interface access lists and NTP access lists. and restrict NTP transactions to that server alone. access-list 120 permit tcp host 14. The NTP source interface is the network connection from which the NTP control messages will be sent. Use the source qualifier to bind the NTP service to the loopback interface. To find out the main NTP servers on the wide-area network you plan to join.2. South(config)# interface eth0/0 South(config-if)# no ntp disable South(config-if)# exit South(config)# ntp server 14.141 eq ntp NTP access lists can be used to impose fine-grained access control on NTP servers.11. # master (unsynced). and how to check that the NTP association is working. For more information about access lists consult Section 4.250 source loopback0 South(config)# exit South# ! wait one minute or so..250 eq ntp host 14.9. South# show ntp associations address ref clock st when poll reach delay offset *~14. The example below shows how to configure the router South to use the router Central as its NTP server.2. consult the network administrator. .250. clients. End with CNTL/Z. one per line.2. Version 1. designate one or more NTP servers. + selected.0 -0. and peers.9. ~configured South# show clock detail 09:30:08.25 * master (synced). check the Cisco IOS documentation for details.203. second. simply designate one or more NTP servers.141 eq ntp access-list 120 permit udp host 14. South# config t Enter configuration commands.9 9 11 512 377 2.250 If you use NTP.2.3. To add an NTP server use the command ntp server. There are two steps to configuring a Cisco router to be a simple NTP client: first. The example below shows access list entries that permit NTP traffic between router South’s loopback0 interface and a designated address of 14. A full explanation of NTP access control is outside the scope of this guide.candidate. The example below shows how to set up an NTP server. then your interface access lists should be configured to permit the NTP protocol (TCP port 123 and UDP port 123) only for designated NTP participants.9.11.

as shown here.candidate.2.Router Security Configuration Guide South# config t Enter configuration commands. 26. To restrict the set of hosts to which your router will provide NTP service. 150 Version 1.15. a Cisco router configured with one or more NTP servers or peers will act as an NTP server. Configuration Sample The configuration command listing below shows the configuration commands for a router with console logging.9.9. one per line.2.9 9 11 512 377 2. # master (unsynced).2. one per line. Configuring NTP Authentication Cisco IOS supports authenticated NTP.250 key 1 source loopback0 South(config)# exit Note that configuring NTP authentication. buffered logging. The host receiving the log messages is 14.9. End with CNTL/Z.25 * master (synced).9.1c . and authenticated network time synchronization. . does not prevent a router from responding to NTP queries from other network hosts. Unless your network is responsible for providing time service to other networks.2. South(config)# ntp authenticate South(config)# ntp authentication-key 1 md5 router South(config)# ntp trusted-key 1 South(config)# ntp server 14. End with CNTL/Z.250 South(config)# access-list 21 deny any South(config)# ntp access-group peer 21 South(config)# exit South# show ntp associations address ref clock st when poll reach delay offset *~14.250. ~configured South# By default. Central# config t Enter configuration commands.6. South(config)# ntp server 14. and the time server is 14. the description below shows how to set up authentication for an Cisco router so that it can use a designated NTP server that uses authentication. use an access list.250 source loopback0 South(config)# access-list 21 permit host 14. The example below shows how to disable NTP server facilities on an interface. Setting up such a community is outside the scope of this guide. Central(config)# interface eth 0/2 Central(config-if)# ntp disable Central(config-if)# end Central# End with CNTL/Z. one per line. This sample is formatted as it would appear in a configuration text file stored on a host for download to the router South. you should disable NTP on all external interfaces. South# config t Enter configuration commands. + selected. syslog logging. which uses pre-placed keys to establish a trusted community of NTP servers and peers.2.0 -0.

or object. the information on each device is organized in a virtual store called a Management Information Base (MIB).Advanced Security Services ! turn on timestamps for log entries service timestamps log datetime msec localtime show-timezone ! setting logging levels and syslog parameters logging console notifications logging monitor debug logging buffered 16000 informational logging facility local6 logging source-interface loopback0 logging 14.2. values for the entire branch are returned. where the dots separate branches in a MIB tree.3.9. An in depth discussion of SNMP data organization is outside the scope of this guide. A MIB is a hierarchical. in a MIB is referenced by an object identifier (OID).sourceforge. for more information consult [7].net/ .2.9. A free and usable implementation of an SNMP agent and manager may be obtained from the NETSNMP home page (http://net-snmp. If the referenced value is not a bottom leaf of the tree.5. SNMP may also be used by an entity on the network to send alerts indicating Version 1.2.1c 151 . the router is always the agent. An OID is a unique.6 logging on ! a tiny access list to permit access only for Central access-list 21 permit 14. SNMP may be used to query the status of or set the values of network components. SNMP is the transport protocol used to share and change information between MIBs. A software application on a PC or workstation normally acts as the manager. Each piece of data. numerical name. tree-like structure used to store a virtual database of network management information. In the case of Cisco routers. the SNMPv3 agent used in the creation of this section). one may use the OID or the actual name of each branch (separated by dots).250 key 1 source loopback0 4. When requesting the value of an object. An SNMP agent device maintains information and makes it accessible to managers.9. Security for the Simple Network Management Protocol (SNMP) The Simple Network Management Protocol (SNMP) supports a connection between two entities that communicate with each other: the manager and the managed entity.NET-SNMP is the successor to ucd-snmp. dotted. the agent.250 access-list 21 deny any ! designate Central as our sole NTP server with authentication ntp authentication-key 1 md5 LTGTR-769015 ntp authenticate ntp trusted-key 1 ntp access-group peer 21 ntp server 14.

This string is sent across the network in clear text. disable SNMP altogether. They are described in the table below. it became widely used. conflict arose between competing proposal advocates and no security standard was agreed upon. DES 56-bit encryption was not supported in the versions of IOS used for preparation of this section (12. Proposals were put forth to integrate security (as well as more functionality) into later versions of the protocol. consult the Cisco security advisory “Malformed SNMP Message-Handling Vulnerabilities” [9]. SNMPv3 defines three levels of security. Consequently strong security was left out of SNMPv2c. IOS version 11. Unfortunately. it was originally intended to be a short-term solution for (remotely) managing networks. 152 Version 1. SNMPv3 was developed specifically with strong security in mind. For more information. those that do not have a direct impact on security are not covered. If your IOS release is one of the vulnerable ones (and virtually every IOS prior to February 2002 is) then you should either upgrade your IOS (recommended). serious SNMP vulnerabilities were disclosed that affected Cisco routers and many other network devices. There are currently three versions of SNMP: SNMPv1. Cisco IOS supports a large number of SNMP-related commands. As such.0 supports all three security levels.0(5)). SNMPv2 relies on the same mechanism for access control to the MIB. However. SNMP Vulnerability In early 2002. However. Table 4-4: SNMPv3 Security Security Level Authentication Username sent in the clear HMAC-MD5 or HMAC-SHA HMAC-MD5 or HMAC-SHA Encryption None None DES (56-bit) SNMPv3 noAuthNoPriv authNoPriv authPriv The Cisco documentation indicates that IOS 12. IOS versions 12. SNMP Security When SNMPv1 was developed. since it was the only network management protocol available at the time. SNMPv1 and SNMPv2c have weak security.0 and later support all three versions of SNMP. SNMPv1 uses a community string to limit access to the MIB.3 supports SNMPv1 and SNMPv2c.1c .Router Security Configuration Guide problems. it was developed quickly and strong security was not a requirement. SNMPv2c and SNMPv3. or take other protective measures. This section will give a brief overview of SNMP security and will detail how to enable SNMP more securely. In the late 1990s.0(7) and 12.

SNMP logging must also be enabled (see section 4. SNMPv3 A Cisco router capable of running SNMPv3 allows for more security measures to be applied. Unfortunately. anybody who knows this community string has access to essentially the entire MIB. there is no other alternative when enabling SNMP. So. While there is some mention of enhanced security options (for SNMPv2c) in the Cisco documentation. The rest of this section focuses on SNMPv3.162.0. This string is stored on the router in clear text and will be sent across the network in the clear. on Cisco IOS version 11. as shown below.Advanced Security Services Configuring SNMP . End with CNTL/Z East(config)# snmp-server community publicstring East(config)# snmp-server host 14.6. East# config t Enter configuration commands. 0 sent. East# Running these basic commands by themselves is not very secure.2.1c 153 . SNMPv3 has been implemented and provides more security features. It is a good idea to disable the public community string.6. these commands have been disabled.1). one per line.5.2. In order to enable SNMP a default community string must be set. It is a good idea to run the show snmp command to display the SNMP status and statistics. 0/10.6. there are some basic commands you must run to enable SNMP. However.3) needs to be created to limit machine access to the router Version 1. 0 dropped. Then an access control list (see Section 4. in version 12.6 traps public East(config)# exit East# show snmp Chassis: east Contact: John Doe Location: Headquarters 0 SNMP packets input 0 Bad SNMP version errors 0 Unknown community name 0 Illegal operation for community name supplied 0 Encoding errors 0 Number of requested variables 0 Number of altered variables 0 Get-request PDUs 0 Get-next PDUs 0 Set-request PDUs 0 SNMP packets output 0 Too big errors (Maximum packet size 2048) 0 No such name errors 0 Bad values errors 0 General errors 0 Response PDUs 0 Trap PDUs SNMP logging: enabled Logging to 14.Getting Started In both IOS versions 11 and 12.3 (which implements SNMPv1 and SNMPv2c).

1c . users and MIB views.Router Security Configuration Guide (through SNMP).6 East(config)# exit After these commands. East# config t Enter configuration commands. Limited access may be given to the MIB by defining groups. is disabled.6. one per line. No community string is required. This group has read access to the “sysonly” view. The second step is to add users to the group. First. the version of SNMP and the security model desired. A MIB view defines a portion of the MIB that a user or group may see/modify provided they have the appropriate credentials. A better method to allow access to the MIB is to use strict controls. The following example defines a non-privileged user. End with CNTL/Z East(config)# snmp-server group publicUser v3 noauth read sysonly East(config)# snmp-server user jdoe publicUser v3 East(config)# snmp-server view sysonly system included East(config)# exit East# East# show snmp group groupname: publicUser security model:v3 noauth readview :sysonly writeview: <no writeview specified> notifyview: <no notifyview specified> row status: active East# East# show snmp user User name: jdoe Engine ID: 00000009020000500F033680 storage-type: nonvolatile active East# East# show snmp view sysonly system . “jdoe”. one per line. If this MIB view is not specified the default is to have access to basically the whole MIB. who is a member of the “publicUser” group. Then a MIB view should be defined to either include specific MIB branches or exclude specific MIB branches. SNMP is still enabled but no one has access to the MIB because the community string.included nonvolatile active East# 154 Version 1. The following example also introduces two new commands used to verify that the new groups and users have been added correctly. which is the “system” branch of the MIB. This branch contains useful information and is beneficial for users to have access to. which solely defined access to the MIB. as well as the access to that view may also be defined. a group must be defined by specifying a group name. instead authentication is based on the user name. East# config t Enter configuration commands. This is an example of a noAuthNoPriv security model. Following is an example that does this.2. A specific SNMP MIB view. End with CNTL/Z East(config)# no snmp-server community publicstring East(config)# ! create access list to use later East(config)# access-list 20 permit 14. More than one machine may be added on the access-list.

follow the guidance below. In this example. the routing tables are not accessible remotely. Likewise. Overview of RMON Remote Monitoring (RMON). one per line.2). 4.1c 155 . it should be disabled. members of the administrator group have restricted read and write access. Otherwise. a group must be defined. groups and views must be defined to control access to the MIB.ipAddrTable excl East(config)# snmp-server view adminview ip. It provides the capability of monitoring and analyzing traffic – data to and from network devices on distributed network segments. The RMON standard was originally developed by the Internet Engineering Task Force (IETF) to provide proactive monitoring and analysis of Version 1. Then users must be added to the group with a password string. it is always a good idea to use the show commands to verify the new settings. This string may be hashed using MD5 or SHA. A MIB view may be defined by more than one included/excluded statement to restrict the view to the appropriate MIB branches. it can be disabled by disabling SNMP (see Section 4. The following example defines a privileged user. Security for Remote Monitoring (RMON) This sub-section describes RMON and security issues related to it.4. RMON is based on SNMP. The steps to support this security model are similar to the steps in supporting the noAuthNoPriv model. it also provides a potentially dangerous conduit into a network. users. This security model uses MD5 or SHA to hash the community string. while not shown. User “root” is a member of the “administrator” group. This makes it harder to compromise the community string. Access-lists. is an extension of SNMP. First. the routing tables are not permitted to be modified remotely. If you are not using RMON. End with CNTL/Z East(config)# snmp-server group administrator v3 auth read adminview write adminview East(config)# snmp-server user root administrator v3 auth md5 “secret” access 20 East(config)# snmp-server view adminview internet included East(config)# snmp-server view adminview ip. his community string “secret” will be hashed and then sent across the network. While SNMP is helpful because it allows an administrator to remotely configure the router.5. “root” who uses MD5 for authentication. This means that when user “root” tries to access/modify MIB data. defined by the view “adminview”.ipRouteTable excl East(config)# exit The examples above showed some basic rules that should be followed when configuring SNMP on a router. even if the community string is somehow compromised. Then the MIB view is defined. So. East# config t Enter configuration commands.Advanced Security Services The more secure model implemented is authNoPriv. This view gives access to all parts of the MIB except the branches that display routing information. Of course. to the MIB.

collect and log information in all nine groups. and a falling threshold. However. Implementation of RMON in Cisco Routers The Cisco IOS versions installed in most Cisco routers. If a data sample crosses a defined threshold. implement a small sub-section of the RMON1 agent standard. Packet Capture. Statistics.1c . the RMON agent implements the Alarm and Event groups only. referred to as RMON2. RMON specifications are defined in the basic RMON standard. The basic IOS RMON agent supports the Alarm and Event groups.0. The alarm group periodically samples statistics from variables and compares them to thresholds configured on the agent. Some network monitoring probes may require a read/write community string in order to communicate with the agent. The implementation of the rising 156 Version 1. basically RMON1. The network management station will record details about all configured events triggered on the monitored router. beginning with IOS 11. Matrix. the RMON agent fires an event. this document covers only those features and security concerns applicable to the most common IOS releases. referred to as RMON1 and in the extended version. Since the RMON option is an add-on enhancement to the Cisco router’s IOS.5. The configuration of the alarm group is dependent on a previously configured RMON event. RFC 1757. then enable SNMP traps on the router (see Section 4.Router Security Configuration Guide traffic data on distributed LAN segments. the polling period. a read/write community string is highly discouraged (see Section 4. In order to enable RMON on the Cisco routers. IOS images ordered with the explicit RMON option.2). The configured parameters of an alarm identify a SNMP MIB variable to monitor. The RMON Management Information Base (MIB) defined in RFC 1757 is a standard method for monitoring basic operations of network devices on LAN segments by providing interoperability between SNMP management stations and RMON monitoring agents. Filters. If the agent installed on the router does not include the explicit RMON option. Host. HostTopN. a network administrator uses applications such as NetScout Manager Plus. Protocol analyzers or RMON probes add enhanced monitoring capability of RMON agents by passively collecting data packets on the monitored LAN segment. RMON1 is widely implemented in most data communication devices. and the Event Groups. The probe communicates the data collected to a Network Management Station via SNMP. As a security precaution.2). RMON1 collects current and historical traffic statistics up to the MAC-layer of the OSI model. Alarm.1 on up to IOS 12. History. logs a message or generates a trap and transmits it to the Network Management station. The event fired. RFC 2021. RMON2 provides traffic-level statistics plus finer granularity of network behavior from the network to the application layers of the OSI model. if the network architecture includes a deployed SNMP infrastructure and network management station. Optivity LAN. a rising threshold with the associated event. In addition. or HP OpenView to process and display the RMON results in graphical or report form. a Read Only community string is required when configuring the standard SNMP agent. On the network management station.

1 30 delta rising-threshold 40 1 falling-threshold 0 owner rscg Central(config)# exit Central# show rmon alarms Alarm 1 is active.13.13. An important requirement that is easily overlooked is the instance number of the monitored variable.1c 157 . Variables included in the SNMP table format will have an instance number equivalent to the entry number of the table. However. Central(config)# rmon alarm 1 ifEntry. counter. etc. All other elementary data variables should have an instance number of ‘0’. The basic IOS RMON agent supports the following commands: show rmon alarms show rmon events rmon event number [log] [trap community] [description string] [owner string] Display information on alarms configured Display information on events configured Configure an RMON event rmon alarm number MIB-object Configure an RMON alarm interval {delta | absolute} rising-threshold value [event-number] falling-threshold value [event-number] [owner string] The first two commands display information on configured RMON facilities. the following command defines an alarm configured on a member of the MIB II interfaces table. .Advanced Security Services and the falling thresholds of an alarm are dependent on the previous configuration of an associated event. last value was 3 Rising threshold is 40. assigned to event 0 On startup enable rising or falling alarm Alarm 2 is active. For example. Use the rmon event command to provide a description of an event and specifies whether a message is logged or a trap is generated. Use the rmon alarm command to designate the actual MIB variable monitored on the Cisco router. Alarms may define any SNMP MIB variable that has an elementary data type such as integer. gauge. assigned to event 1 Falling threshold is 0. owned by config . RMON alarms provide an excellent tool for monitoring the network interfaces supported by the router. one per line. It is acceptable to use the Object Identifier (OID) or the qualified variable name that resolves to its OID. Central# Version 1. there are several limitations on the type of SNMP variables RMON is capable of monitoring. ifTable: Central# config t Enter configuration commands. timeticks. All monitored objects must include an instance number of the monitored variable. End with CNTL/Z. owned by rscg Monitors ifEntry. The MIB object monitored must also resolve to an ASN.1 notation.1 every 30 second(s) Taking delta samples.

.113-4. It is important to note that most Cisco updates can only be accomplished by replacing the IOS software running on the router. booted via flash cisco 3640 (R4700) processor with 28672K/4096K bytes of memory. identifies variable ifInDiscards. 158 Version 1. Central> show version IOS(tm) 3600 Software (C3640-I-M). Inc. RMON may be very data intensive depending on the number of monitored variables and the length of the sampling period. . there is no facility for amending or patching installed IOS software. then change the sampling period to a longer time (e. If the amount of traffic generated by RMON seems to be too high. simply add the two parts of the RAM size rating: this router has 32MB of RAM. 8192K bytes of processor board System flash (Read/Write) . System image file is "flash:c3640-i-mz. unknown or unsupported protocols. router model. Central> The underlined portions of the transcript are the software version. 30 seconds to 60 seconds). . An upgrade can be beneficial for security.Router Security Configuration Guide The interface entry.3(4)T1. The agent monitors increases of forty discarded packets or more starting from the last value sampled. Performing Cisco IOS Software Updates This sub-section outlines the motivations and procedures for upgrading the system software on a Cisco router. . RAM size. To determine the current software release running on a router. Alarm number 1 defines a sampling period of every 30 seconds for the number of discarded packets inbound to the Ethernet interface stored at table entry 1 or instance 1.5. and flash memory size. ifEntry.1c . .13. RELEASE (fc1) Copyright (c) 1986-1998 by cisco Systems.g. respectively. but if done improperly it can leave a router vulnerable. 4. the number of inbound packets discarded. To compute the total RAM on the router. and identify the version and memory size as shown below. etc. input and output discarded packets.1. input and output errors. This section also presents information about backing out of an upgrade. It is important to know the router model and memory sizes before attempting to obtain a software upgrade.T1". A router’s RMON agent can be very useful for monitoring the number of checksum. Version 11. use the command show version.5.

It is very important to read the release notes for a new release and test it carefully before installing it for operational use. Software updates entail substantial costs. Print the summary web page. 1. you will be given the length of the release in bytes. To support new features – Cisco has added new operational and security features to each new IOS release. Obtaining Updates Cisco makes software updates available through a variety of purchase and maintenance mechanisms. Third. router memory size. the router must be out of service for at least a short time during the installation process. For more information about Cisco IOS release types. you must take many factors into account: feature availability. or to enforce your local security policy. it is best to check the length after downloading. they might be broken or simply unsupported. 2. to ensure that the new software can fully support the router functions your network needs. If you need one or more of these features to support your network. depending on router model and other factors. which will include the length. and again afterwards. The logistics of purchasing updates is beyond the scope of this document. To improve performance – you might need an upgrade to support new hardware or hardware features. a new release may degrade performance. Weigh the benefits of upgrading against the risks before you start. either by implementing new features or by reducing available free memory. you can download updates from the Software Center on the Cisco web site. release status. If the performance of your router is critical. be prepared to back out if the performance has suffered. some features may not work in a newer release. cost. for the IOS image you’ve Version 1. Whenever you download Cisco IOS software (often called an IOS “image”). During the software selection and download sequence at Cisco’s web site.3. see Section 8. First.Advanced Security Services Motivations for Updating Router Software Installing an IOS update entails inconvenience and the risk of disruption of service. and bug history. then measure the performance before upgrading. then do the upgrade. then it makes sense to upgrade. If the performance benefit is greater than the cost of upgrading. Deciding which update to pick is a complex topic.1c 159 . To fix known vulnerabilities – when security vulnerabilities are found in Cisco IOS products. 3. Second. one solution may be to upgrade to a later edition of the IOS software. The list below describes some good reasons for installing an update. If you have a maintenance agreement with Cisco. the minimum downtime will range from at least a minute to several minutes.

use SCP or FTP for performing Cisco upgrades. 160 Version 1. If possible. a secure file copy protocol based on SSH. When possible. below. then FTP will probably not be available. and inform the user community as needed. then disable it again. it is prudent to configure operational routers with as much memory as they can hold. Most newer IOS releases support SCP. or a similar checksum utility. To use SCP. 4. Check your TFTP. Ensure that you have enough memory. If the length or hash of your file differ from the summary page.1c . or FTP configuration.0 supports FTP. Enable TFTP only for the update sequence. Copy the new Cisco IOS software into the server’s download directory. Installing an update imposes a minimum downtime. Be sure that you are familiar with all the IOS commands involved. (Often. a major upgrade will require more memory. (If the router to be upgraded is running IOS 11. Use the commands show version and show flash to check the amount of memory your router has. and should not be running in a secure network. discard the file and download it again.3 or earlier. For more information. 1. 3. Review the entire procedure before you start. you will need to have an SSH server running on a host in your network. check the length against the printed page.Router Security Configuration Guide selected.) While TFTP is supported by all IOS versions. most earlier releases do not.) 2. Schedule your upgrade ahead of time. Every Cisco IOS release has minimum memory requirements. Cisco IOS 12. and may impose much longer downtime (up to half an hour if things go wrong and you have to back out). Make sure that the TFTP or FTP server is correctly set up for both upload and download. Do not install an update unless the router to be upgraded satisfies the memory requirements for both RAM and Flash. it is not a secure service. Router software updates are usually performed using TFTP or FTP. use the md5sum command. The summary page will also include the MD5 hash value for the IOS image. RCP. to check the hash value. consult [12]. because many Cisco routers are configured with just enough memory to run the IOS version pre-installed at the time of purchase. After downloading the IOS binary file. Cisco routers have two fundamental kinds of memory: RAM and Flash. Schedule your downtime. Before You Perform the Update Check all the items below before installing a new IOS image on your router. Read the entire upgrade procedure.

take advantage of that to perform the update without disrupting service. and back up the current IOS software. Central# copy flash: tftp: or Central# copy flash: ftp: You will need to supply a file name and the IP address or host name of the TFTP or FTP server host. steps 1-3. and ensure that you can restore your previous IOS version if necessary. Check the current IOS version number and flash contents with the commands show version and show flash. 2. Enable privileges. If a redundant router or a hot spare is available. set an FTP username and password that will work on your server (otherwise. Copy the router’s current IOS image to the server using the copy command as shown below. If you are planning to use FTP. abandon the update and check the server configuration before trying again. it is safest to replace a router and take it offline for update. confirm the current IOS and boot version. If the router to be upgraded is a border router. The steps are described below. The sequence is very conservative. 0. Update Procedure This section presents a suggested sequence of steps for installing Cisco IOS software. The console will show important status messages in the later steps of the installation that would not be visible otherwise.1c 161 . and test. involves copying the running IOS software and configuration onto the TFTP server host for safekeeping. by following it you can avoid mishaps. Shut down external interfaces. involves loading the new software. It is best to perform router updates from the system console rather than from a network login. Central# config t Central(config)# interface eth 0/0 Central(config-if)# shutdown Central(config-if)# end Version 1. do not proceed. then disable the outside network interfaces using the shutdown command. step 4. The test phase. steps 56. install.Advanced Security Services If possible. followed by a console transcript of a successful update. The sequence has three phases: backup. For example: Central(config)# ip ftp username uploader Central(config)# ip ftp password Y3ti4ttack! Central(config)# end 1. If this step fails. The backup phase. The install phase. involves checking that the new software is running the old configuration successfully. make a record of them. Log in on the router console. the router will attempt to log in as user “anonymous”).

Central# reload Proceed with reload? [confirm] y 5. Load the new software. If this step fails. Central# copy tftp: flash: or Central# copy ftp: flash: On some Cisco routers. . (If you have several IOS images in flash. Central# show ip interface . Confirm the new IOS version and boot image. if it does not. Copy your current startup configuration to your TFTP or FTP server using the copy command as shown below.1c . the flash will be erased automatically during this step. Central# 162 Version 1. On some Cisco routers. Central# copy startup-config tftp: or Central# copy startup-config ftp: You must supply the IP address or host name of the server host. Copy the new IOS software from the TFTP or FTP server to the flash memory of the router. your router may automatically reboot.) If this copy succeeds.0(9). . if asked whether to erase the flash. Use the copy command as follows. answer yes. . you can specify which one to use at boot using the boot system command. Back up the current running configuration. enable privileges and confirm the configuration status with show running-config. . re-establish the connection at this point and check the IOS version and boot image with show version. Check the status of the interfaces.Router Security Configuration Guide 3. Central# show running-config . Version 12. and check that the access lists and static routes are still present. . If you are performing the update over a network connection. If you performed steps 1 through 4 over a network connection. it is possible to store several IOS releases in flash memory and select which one to run. do not proceed. Then. your connection will be broken at this point. 4. abandon the update and check your server configuration before trying again. . RELEASE SOFTWARE . then reboot it manually using the command reload. Central# show version Cisco Internetwork Operating System Software IOS(tm) 1600 Software (C1600-SY56I-M). Watch the boot messages on the router console to confirm the new IOS software version and boot image.

T1' (file # 1).113-4. for some older Cisco router models. Note that. South(config)# interface ethernet0/1 South(config-if)# shutdown South(config-if)# exit South(config)# exit South# Version 1.9. they should have come back up as part of the reload in step 4. Central# config t Central(config)# interface eth 0/0 Central(config)# no shutdown Central(config)# end Depending on network speed and router model.T1. this procedure may take about 5-20 minutes.bak'? [yes/no]yes !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Upload to server done Flash device copy took 00:00:19 [hh:mm:ss] South# config t Enter configuration commands. If you shut down your router’s external interfaces in step 2.T1 Destination file name [c3640-i-mz. RELEASE SOFTWARE (fc1) . Bring up external interfaces. Consult the release notes for the particular router for details.Advanced Security Services 6..0(5). Transcript of a Successful Update Procedure The recorded transcript below shows an upgrade of a Cisco 3640 router from IOS 11.3(4)T1. If the second command in step 5 showed that they did not come back up. additional hardware-specific steps may be needed.113-4.T1]? c3640-i-mz-113-4. OK Copy 'c3640-i-mz.T1' from Flash to server as 'c3640-i-mz-113-4.. if necessary. South> show version Cisco Internetwork Operating System Software IOS(tm) 3600 Software (C3640-I-M). 8388608 total] Address or name of remote host [14. Version 11.113-4.113-4.T1.6]? 14.bak Verifying checksum for 'c3640-i-mz. 8388608 total] 8192K bytes of processor board System flash (Read/Write) South> enable Password: South# copy flash: tftp System flash directory: File Length Name/status 1 3208548 c3640-i-mz.3(4) to 12.2.113-4. .9.6 Source file name? c3640-i-mz.2. 5179996 available. End with CNTL/Z.1c 163 .T1 [3208612 bytes used.113-4.T1 [3208612 bytes used. 5179996 available. one per line. then bring them back up now using the command no shutdown. South>show flash System flash directory: File Length Name/status 1 3208548 c3640-i-mz.

8388608 total] Address or name of remote host [255.2.6? [confirm] Building configuration.bak Write file south-config.255.6 (via Ethernet0/0): ! [OK] Erase flash device before writing? [confirm] Flash contains files.bak on host 14.erased Loading c3640-ik2o3s-mz_120-5_T1.9. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee ..2.T1 [3208612 bytes used.6 (via Ethernet0/0): !!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! [OK .bin' from server as 'c3640-ik2o3s-mz_120-5_T1.6 Name of configuration file to write [south-confg]? south-config.1c . size: 0x74d170 Self decompressing the image : ############################################# [OK] ..9.9. .1(19)AA.2.bin Destination file name [c3640-ik2o3s-mz_120-5_T1. Version 11.bin' into Flash WITH erase? [yes/no]yes Erasing device.bin]? c3640-ik2o3s-mz_1205_T1. entry point: 0x80008000... OK (0xDC3B) Flash device copy took 00:00:50 [hh:mm:ss] South# reload System configuration has been modified. South> South> show ip interface brief Interface IP-Address OK? Method Status Protocol Ethernet0/0 14...9.bin' on 14..7656076/8388608 bytes] Verifying checksum.2.bin from 14.bin Accessing file 'c3640-ik2o3s-mz_120-5_T1. 164 Version 1. EARLY DEPLOYMENT SOFTWARE (fc1) Copyright (c) 1998 by cisco Systems.250 YES NVRAM up up Ethernet0/2 unassigned YES NVRAM administratively down down Ethernet0/3 unassigned YES NVRAM administratively down down South> enable Password: South# show running-config Building configuration.2.. Loading c3640-ik2o3s-mz_120-5_T1. Are you sure you want to erase? [confirm] Copy 'c3640-ik2o3s-mz_120-5_T1..6 Source file name? c3640-ik2o3s-mz_120-5_T1. C3600 processor with 32768 Kbytes of main memory Main memory is configured to 64 bit mode with parity disabled program load complete.64 YES NVRAM up up Ethernet0/1 14.113-4.255]? 14..2.9.Router Security Configuration Guide South# copy running-config tftp Remote host []? 14.bak !! [OK] South# copy tftp flash System flash directory: File Length Name/status 1 3208548 c3640-i-mz.. 5179996 available.9.6. Writing south-config.2.bin from 14.255. Save? [yes/no]: no Proceed with reload? [confirm] y %SYS-5-RELOAD: Reload requested System Bootstrap..2.9.10. Inc.

2 to 12. copy tftp flash reload ! Optional. If the router supports SCP. Central# . but the simplest is to ensure that the TFTP traffic does not traverse hostile networks. do not leave TFTP enabled on your host.3). Second. South# exit Backing Out an Update If functional testing reveals a problem with your router after an upgrade.0 . Simply follow the procedure described above. Central# . this section attempts to address them. from 11. if you follow the installation procedure outlined above. whenever you make any kind of backup copy of a router configuration. always use it instead of TFTP.g. it is critical that you protect the TFTP transaction and server from potential attackers. your stored configuration might not work correctly when you fall back to the older version. load the backup copy of the old IOS software. Central# Central# . you may need to return to your old IOS version.1c 165 . Because TFTP provides no security. Some of these settings can affect your router’s security. use a different name than you used during the upgrade procedure. Third. some newer versions offer services not present in older versions (see Section 8. Also. First. Note that. always turn it off immediately after you finish the installation procedure. Version 1. . In step 3. restore old configuration copy tftp running-config Additional Security Concerns There are several security issues surrounding upgrades. if you have upgraded from one IOS major version to another (e. starting with step 2. restore the backup copy of the configuration that you saved during the upgrade procedure step 1. . The simplest approach to mitigating this risk is to change the enable secret immediately after installation (see Section 4.0). There are several approaches to doing this. you transmit a copy of your router configuration to a TFTP server. . In step 4. many default settings differ between various IOS releases.6).1) or to use a centralized authentication server (see Section 4. In that case. . you may be exposing your encrypted passwords to disclosure. Also.Advanced Security Services Current configuration: ! version 12.

In the example below. keep input/output statistics. and they offer a great deal of visibility into the operation of the router. The messages are shown in the order they occurred. There are literally hundreds of such queries available. and are useful for understanding a Cisco router’s disposition in a typical TCP/IP network. identify sources and causes. 1.Router Security Configuration Guide 4. Diagnosing and Debugging Router Operation Effective logging and SNMP help an administrator to stay aware of their routers’ status and operational condition. an unauthorized attempt to telnet to the router itself has been logged by access list 131. East# show logging 166 Version 1. and validate repairs.2. and some is particular to each interface. These commands typically produce a correspondingly great deal of output. consult Section 4. see [2] and [7]. so recent messages are at the end of the listing.6. For more discussion of logging. These commands can also be used to help verify that security measures are in force. Router Status and Configuration Commands Each of the items below describes a single status query. and some other facilities.5. • Debugging commands – Virtually every IOS facility and protocol has associated debugging commands. for a discussion of some other useful ones. or when a network is under attack.1c . Techniques for troubleshooting and debugging routers could (and do) fill entire books. Cisco IOS diagnostic and debug facilities can be used to get vital information. Testing and validation are covered in Section 6. This short sub-section describes some of the most useful techniques for IOS 11. Viewing the current log – To view the current buffered log messages. so the first few messages put into the log reflect startup activity. The ones listed here are commonly used for simple troubleshooting.5. When a problem occurs. even on the simplest Cisco routers. The buffered log messages are cleared when the router reboots.3 and later. use the command show logging. so use them sparingly. some of this information is global to the whole router. • Router throughput and traffic commands – Each interface. The output consists of two parts: a summary of the current logging configuration. and the log messages. The techniques fall into three groups: • Router status and configuration commands – These commands display information about the settings and tables held by the router. There are IOS commands to display these statistics that can be used to detect problems.

64.6.0 [110/10] via 14.9. 00:01:05.0/16 [110/11] via 14.0. Viewing the current route table – To view the current route table.2.0/0 [110/3] via 14.1.0/8 [110/14] via 14.0.0. Ethernet0/0 14. changed state to up .0. Mar 3 12:51:52 EST: %SEC-6-IPACCESSLOGP: list 131 denied tcp 172.0 [120/1] via 14. Each route code identifies how one route joined the table. the first few log messages will include the time since boot instead of the correct time.9.1.0/8 [110/12] via 14.6. Security Services Syslog logging: enabled (0 messages dropped. A very important part of reviewing the route table is checking the route codes and checking the destination gateway.250.1. 33 message lines logged Logging to 14. or were added by a configured routing protocol (codes R. 1d01h. Figure 4-11 shows how to interpret the output of show ip route. because the messages are generated before NTP has synchronized.2.2. see Section 4. Ethernet0/1 0.250(47746) -> 0. Ethernet0/0 14.0 O IA O IA O C O E2 C R O*E2 7. Gateway of last resort is 14.1. Ethernet0/0 172.0. Ethernet 0/0 Route codes Destination gateways Figure 4-11: Interpreting a Route Table Listing Version 1.0. Depending on the size of the network and the kinds of routing protocols used.0.0.1. 2d18h. use the command show ip route. O. 1 packet East# Note: log messages should always include the time of the event.9. or were added as static routes (code S).0(23).2. Note that the route table listing on an operational router will often be much longer than this sample.250.250 to network 0.0 is directly connected.2.1.250. 56 message lines logged Trap logging: level informational. Ethernet0/0 14. Ethernet0/0 7.1. the destination gateway is simply the next hop on that route.101.17. 2d19h.1. and others. 2d18h.250. Check the route codes to make sure that all the routes joined the table either directly (code C).1c 167 . 32 message lines logged Buffer logging: level debugging.0. 33 message lines logged Log Buffer (16000 bytes): 00:00:17: %LINK-3-UPDOWN: Interface Ethernet0.18. Ethernet0/1 14.10. . In a router using NTP.1.0/16 is directly connected. 1d13h.0. 56 message lines logged Monitor logging: level debugging. this list may be very large.4). 0 flushes) Console logging: level debugging.1.20.

0f03.1.250 1 FULL/DR 00:00:38 14.3680 0010. Viewing the routing protocols in use – The command show ip protocol gives a verbose listing of the route update mechanisms currently used on the router. and DNS.baa0 Type Interface ARPA Eth0/1 ARPA Eth0/0 ARPA Eth0/1 ARPA Eth0/0 5. Central# Line 0 con 130 vty *131 vty Central# show users User 0 jsmith 0 andrew 1 neal Host(s) Idle Location idle 00:00:56 idle 00:01:02 14.64 43 Internet 14.2. The example below shows the IP routing protocol summary and (abbreviated) output for a useful OSPF status command.250 53 . Viewing the logged in users – The command show users displays a list of users that are currently logged in.f3f6 0010.20 1 FULL/DR 00:00:33 14. use the command show arp.7bf9.9.1. and unauthorized devices on a network segment can often be detected by their presence in a router’s address resolution (ARP) table. The output is different for each kind of protocol. Central# show arp Protocol Address Age(min) Internet 14.Router Security Configuration Guide 3. Central# show ip protocol summary Index Process Name 0 connected 1 static 2 ospf 1 3 rip Central# show ip ospf neighbor Neighbor ID Pri State Dead Time Address Interface 14.1.1c . Viewing host name and name lookup information – Cisco IOS uses two mechanisms for mapping between IP addresses and names: locally defined names. In the example output below.2. Locally defined names take 168 Version 1.9. there is one user logged in at the console.2.20 idle 00:00:00 14. mis-connected devices.127a 0050.250 Eth0/0 Central# 4.6 57 Internet for some recommendations. The command show ip protocol summary gives a quick overview.20 Eth0/0 14. To display the ARP table. Central# Hardware Addr 0004.1.1. All of the individual routing protocols also have extensive status commands.2. Viewing the current ARP table – Extraneous devices. . as in the example below.6 6.1.9. see Section 10 Internet 14.acd5. and two are logged in over the network.

9.1 Host Flags east (perm.1.250 YES Ethernet0/2 unassigned YES Ethernet0/3 unassigned YES Central# Method NVRAM NVRAM unset unset Status up up down down Protocol up up down down 8. To display the full status of a line. but can provide valuable clues to the services that the router is actually providing. 14.6 514 14.64 7.Advanced Security Services precedence over DNS names. 6082 Central# In Out Stat TTY 0 0 1 0 0 0 1 0 0 0 1 0 0 0 10 132 Version 1. and typically five virtual line connections. show line aux For a quick look. for instance. use the command show ip interface brief. Viewing currently open UDP sockets – Use the command show ip socket to list the currently open UDP network service sockets on the router.1. Viewing line status – Every Cisco router has at least one physical line connection. OK) south (perm. The example below shows the output for a router running fairly few services.15. slightly abbreviated. the listing will include both active and inactive interfaces.250 520 17 14.2. 9. the console. the telnet vty lines.250 161 17 0.15. OK) Central# Age 4 ** 52 Type IP IP IP Address(es) 36269 14. Central# show ip interf brief Interface IP-Address OK? Ethernet0/0 14. Viewing interface status and configuration – Use the command show ip interface to view a verbose display of the status and configuration of a router’s network interfaces.250 123 17 169 .1.0. The example below shows the brief output format. Central# show host Default domain is not set Name/address lookup uses domain service Name servers are 14. Central# show ip sockets Proto Remote Port Local Port 17 0.1.0 123 OK) central (perm. In all cases.1. Use the command show host to display the DNS configuration and the list of locally defined names.15.0.15. use show line name number. Use the command show line to display a summary of lines available on a router (see Section 14.250 YES Ethernet0/1 14. The output is a little cryptic.0 520 14.

print it. If the configurations are very large and complex.txt Writing central-config. Central(config)# ip ftp password 0 r0ut3rQQ Central(config)# ip ftp user rscg Central(config)# exit Central# copy running-config ftp Address or name of remote host []? 14. The resulting output will typically be fairly long.1 Destination filename [central-confg]? central-config. but TFTP can be used for making archive copies in a very similar manner (see Section 4. one per line.716 secs (474 bytes/sec) Central# In IOS 11. save the command results to a file. Archive a copy of the configuration after any major change.2. Another way to get an archive copy of the running configuration is to use text logging features of Telnet and terminal emulation applications. Because TFTP is insecure. port 123).txt !! 5699 bytes copied in 12. Viewing the current configuration – To view the current running IOS configuration. or on a monthly basis. Use the command show process to list the running processes. The fourth line is the logging client. To review a configuration in depth.5).1c .9. using IOS 12. The third line is the network time service (NTP. 11. and also shorten downtime if the router loses its stored configuration. Normally. use a file comparison tool.3 and earlier. to highlight the differences.Router Security Configuration Guide The first line is the RIP route protocol service (local port 520). 10. Viewing currently running processes – Many IOS services and facilities run as separate IOS processes. To view the saved startup configuration (in NVRAM) use show startup. This can help with problems. these two configurations should be very similar. The output is usually quite long. Check for unwanted processes and services. The example below shows how to save an archive copy of a configuration to an FTP server.0. it should be used with care and disabled when not in use. End with CNTL/Z. such as Unix diff or Windows fc.5. use the command show running. and review the hardcopy. Central# config t Enter configuration commands. 170 Version 1. sending syslog messages to a Unix host (remote port 514). The second line is the SNMP service to a host running an SNMP/RMON management tool (local port 161). FTP is not supported.

7357.0 giants. This will display a comprehensive report on the traffic through all the interfaces. 0 overrun. then clear the counters on a periodic basis. address is 0050. line protocol is up Hardware is AmdP2. 78 collisions. 0 no carrier 0 output buffer failures. use the command show ip traffic. 0 output buffers swapped out Central# If traffic volume monitoring is important for a particular interface. but can reveal certain classes of attacks. The example below shows the output format for a single Ethernet interface. 215 deferred 0 lost carrier. The example below shows how to clear the counters for a single interface. . Last clearing of "show interface" counters 23:20:53 . Clearing the counters sets the traffic volume record back to zero for both input and output. . Central# show interface eth 0/0 Ethernet0/0 is up. . 0 underruns 0 output errors. To view the traffic for a single interface.0 throttles 0 input errors. 3. Most of these commands produce voluminous but clearly formatted output.15. or component failure. 0 late collision. 38371898 bytes.Advanced Security Services Router Throughput and Traffic Commands The commands listed below display various traffic statistics that can be useful in diagnosing router traffic flow. Viewing IP Protocol Statistics – To display a long listing of IP and related protocol traffic statistics. Central# clear counter Eth 0/0 Clear "show interface" counters on this interface [confirm]y Central# 2. . 1. use the command show interface. 0 no buffer Received 800624 broadcasts. The output is quite long.0 runts.1c 171 . Understanding normal network and link traffic loads can be critical for identifying anomalous conditions that are indications of attacks. 0 CRC. 1 interface resets 0 babbles. There are hundreds of traffic and processing status commands in Cisco IOS. misconfiguration. see [7] for more information about some of them. use the very Version 1.1. 0 frame. simply supply that interface name to the command. Viewing the network traffic on a per-interface basis – To view the total traffic for each interface.cbe0 Internet address is 14. 0 ignored 0 input packets with dribble condition detected 480919 packets output. 991606 packets input.250/24 . 103806395 bytes. Viewing SNMP Protocol Statistics – To display the SNMP messages statistics and configuration.

when generated. all activities of that protocol or feature will generate log messages at level 7. Central# show snmp Chassis: Central Contact: Vanessa & Phyl Location: second floor 73 SNMP packets input 0 Bad SNMP version errors 0 Unknown community name 0 Illegal operation for community name supplied 0 Encoding errors 263 Number of requested variables 0 Number of altered variables 10 Get-request PDUs 63 Get-next PDUs 0 Set-request PDUs 73 SNMP packets output 0 Too big errors (Maximum packet size 1500) 2 No such name errors 0 Bad values errors 0 General errors 73 Response PDUs 0 Trap PDUs SNMP logging: disabled Central# The only way to clear these SNMP statistics is to reset the router. Debug messages.9. when using the debug messages for interactive troubleshooting. then the router may have been subjected to an SNMP sweep by an attacker. or feature of the router. facility. The example below shows how to configure debugging and turn on debugging messages for ICMP.1c . The number of messages generated by debugging can often be quite large.2. be sure to configure the buffered log and syslog for level 6 (informational). Central# show users Line User *130 vty 0 rscg Central# config t Host(s) idle Idle Location 00:00:00 14. Therefore. debugging can impose a substantial computational burden. are sent to all log sources configured to receive them. If the output shows any SNMP traffic. and should be used sparingly on operational routers. Each debugging command is associated with a particular service. Router Debug Commands Cisco IOS offers a very extensive suite of debugging commands. and the network does not have an SNMP infrastructure deployed. The example below shows the output for a router with a very modest amount of SNMP traffic.Router Security Configuration Guide simple command show snmp.6 172 Version 1. When debugging is enabled for a particular protocol or feature. Also.

dst 14. RFC 1305. including examining the configuration and operation of a Cisco router. 2000. src 14.M.5.9.6) Central# terminal monitor Central# debug ip icmp ICMP packet debugging is on Central# ! At this point. 1997.2. 1995.9.0 Configuration Fundamentals. 173 . A practical introduction to the concepts and practices for using Cisco routers. [5] Coulibaly. O’Reilly Associates. References [1] Albritton.6 Central# no debug ip icmp ICMP packet debugging is off Central# The Cisco documentation set includes a volume with comprehensive information about the debug facilities and their behavior. D. McGraw-Hill. The sections on “Performing Basic System Management” and “Monitoring the Router and Network” include valuable advice on how to configure basic features and services. M. [2] Cisco IOS 12.2. An excellent introduction to basic IOS operations.2.7. An amazingly detailed book about IOS versions and the IOS release process.159 EST: %SYS-5-CONFIG_I: Configured from console by rscg on vty0 (14.Advanced Security Services Enter configuration commands.6 Mar 3 17:02:13 EST: ICMP: echo reply sent. Cisco Press. Version 1. [4] Mills. the Cisco IOS Debug Command Reference. Cisco Press. The section on “File Management” provides extensive information on downloading updates. End with CNTL/Z. “ping central” was performed on 14. 1999. Central(config)# logging console information Central(config)# logging monitor debug Central(config)# logging buffered information Central(config)# logging trap information Central(config)# exit Central# Mar 3 17:01:58. Cisco IOS Essentials. The specification for NTP version 3. S..250.2. 1999. Cisco IOS Releases: The Complete Reference. “Network Time Protocol (version 3)”. one per line. the version supported by IOS 11 and 12. J. Managing IP Networks with Cisco Routers. Consult this book for information on upgrade paths and compatibility.M. [3] Ballew.

It also shows how to set up an IOS router as an NTP master. G. [12] Stewart. B. and Wright. J. Network Computing. D..1c . Cisco Technical White Papers. A Practical Guide to SNMPv3 and Network Management. including both standard and vendor-specific structures.html This magazine article from 1999 gives a clear and readable overview of NTP. 2002.html This note offers guidance on NTP hierarchy design and tracking performance metrics. A detailed exploration of the SNMP management information April 2002.. This Cisco guide for Internet Service Providers includes a good discussion of IOS upgrades. Cisco ISP Essentials.networkcomputing. A very specific guide to configuring many IOS features securely.. Cisco Press. Cisco Systems. “Getting in Sync: A Look at NTP”. Securing Cisco Routers: Step-by-Step.shtml This Cisco security advisory provides detailed information for dealing with the 2002 SNMP vulnerability on Cisco IOS devices. as well as examples of router status and diagnostic commands. Cisco Security Advisory. Understanding SNMP MIBs. [8] Greene. 1st edition..Best Practices White Paper”. especially for initial set-up of a new router. Prentice Hall. available at: http://www. P. J.. 174 Version 1. E. Includes guidance on using SCP for updates. [10] Shipley.Router Security Configuration Guide [6] Zeltserman. and Smith. available at: http://www. [7] McGinnis. [9] “Malformed SNMP Message-Handling Vulnerabilities”.cisco. SANS Institute. with some configuration suggestions. and Perkins. [11] “Cisco Network Time Protocol . available at: http://www. including good coverage of the SNMP basics and SNMPv3 security features. 1999. Prentice-Hall. An in-depth study of SNMPv3 and its use. May 2002. Cisco Systems. Feb 2002. January 1999.

But the majority of this section will cover using Cisco’s AAA services for controlling administrative access to a router and the security server protocols. authorization. 2. and Kerberos security services provide the facilities required for AAA. Configuration is performed by using method lists as described further below. AAA allows you to base authentication decisions on the router’s local user database. Version 1. This allows for easier management of user accounts and privileges. and Support Mechanisms Cisco’s authentication. and accounting (AAA) services.Advanced Security Services 4. This means that you can no longer access the commands to configure the older protocols (including login local and login commands). Three conditions make using network security servers a good choice: 1. when flexible authorization capabilities are required. AAA also has the ability to control each user’s access to resources and provides additional accounting capabilities beyond the router’s logging facilities. The sub-section below dealing with dial-in users will give an introduction to controlling remote users accessing network resources. the older forms of access control are disabled. selectively. and accounting are particularly important when your organization has many routers and other network devices to manage. and provides additional capabilities for auditing of network service usage. and line passwords.1. Overview. All Cisco IOS releases support the RADIUS and TACACS+ network security services. Centralized authentication. 4. An Internet Service Provider would be a good example of this. when accounting is required. By using AAA along with a security server you can control access to routers and other network services from a centralized location. AAA allows you to employ. authorization. or 3. Cisco provides this security with their authentication. enable.1c 175 . both network security services and security information local to the router. Where the older access control mechanisms dealt almost solely with user authentication. RADIUS.6. AAA is the mechanism Cisco recommends for access control. AAA is designed to allow the administrator to configure its services globally or by line and interface. When AAA services are enabled on a Cisco router. In addition to network security services.6. and accounting services provide critical security functions necessary for providing remote access to routers and network resources. Security for Router Network Access Services Security for Network Access Services deals primarily with controlling remote users who are accessing local resources. and many releases also support Kerberos. Basic Concepts. except Kerberos does not accept accounting records. TACACS+. when there a large number of routers so that centralized administration becomes advantageous. authorization.

6. For more information. if no local usernames are defined then remote administration (via Telnet.2 demonstrates how to set up AAA authentication. Although authentication can be configured without using AAA (see Section 4. it is best to use capable and well-managed network security services as your primary AAA mechanisms. Authentication Authentication is the mechanism for identifying users before allowing access to network components or services. Define one or more method lists for AAA authentication. Configuring AAA authentication entails four basic steps: 1. Configure security server network parameters. etc) will not be possible. and supported token technologies. Kerberos. AAA authentication is set up using method lists. Therefore. Examples in this section will use a subset of the main network diagram as shown in the "Putting It Together" sub-section in 4. When AAA authentication has not been set up the default will use the local username information. AAA authentication provides the means for identifying users through login/password dialogs. In other words. SSH.1). enable passwords and none. This can be done by a combination of named lists and the default list (see the sub-section “Method lists” below for a complete listing). 2.6.2. to use security server protocols or backup authentication methods you must use AAA authentication. see the sub-section on “Method Lists” below. challenge/response mechanisms. but console access will be allowed.Router Security Configuration Guide Note: When using the local user database instead of a network security server. Section 4. Enable AAA (new-model). local username database. 176 Version 1. Apply the method lists to a particular interface or line (optional). TACACS+. in this case. AAA is very limited in its authorization capabilities and provides no mechanism for accounting. For AAA authentication the available methods are RADIUS. and configure local user or line password support only as fallback mechanisms for when the network security services are unavailable. The default method list will be automatically applied to all the lines and interfaces for which a named list was not applied. The authentication method list defines the types of authentication to be performed and the sequence in which to apply them.1c . The following sections will discuss the three main faculties provided by AAA and their supporting concepts. authentication controls the ability of a user or another network component to access a network device or service. 4. 3. Named lists must be applied to the appropriate lines and interfaces. line passwords.

and local methods. Authorization lists are specific to the authorization type which is being defined. and authorization for each user. as described in Section 4. and define the rights for each user. But named method lists. then the security server database will be used to control access to various network components and services as defined by the stored attributes. Authorization is the method used to describe what a user has the right to do once they are authenticated to the router. There is a special case for the console line. PC on the administrator’s desk) to the router. PPP NCPs. configure AAA authentication (since authorization relies on authentication's output). • commands <level> – which controls access to all the commands at the specified privilege level. AAA does not protect the confidentiality of the password during the trip from the remote administration host (e.Advanced Security Services NOTE: The AAA network security protocols each include mechanisms (more or less effective) for protecting the confidentiality of passwords during the exchange between the router and the security server. Version 1.g. Prerequisites for AAA authorization are: enable AAA services. if-authenticated and local. Authorization works by creating a list of attributes which describe what the user is allowed to do. define security servers. if-authenticated. SLIP. if a user has been authenticated when logging into the console line then authorization will not be used (even if configured). and ARA protocols. Authorization includes one-time authorization.) As with authentication. If no authorization list is defined for the authorization type then no authorization will occur for that type.4. Default method lists are applied to all lines and interfaces for that particular authorization type. authorization for each service. Authorization Authorization controls access to system resources. use attribute-value pairs to define a user's rights. AAA authorization types are: • exec – which controls the users ability to run an EXEC shell. as discussed in Section 5. To prevent passwords from being exposed in the clear you must use a secure remote administration with a protocol like SSH or IPSec. The RADIUS and TACACS+ security servers. (The last two methods.6. method lists define what authorization protocols will be used and in what order. Additionally. authorization can only be configured using AAA. Authorization method lists can include RADIUS and TACACS+ security protocols along with Kerberos Instance Maps. must be applied to the interface or line to be invoked. are very limited. After a user logs in and has been identified by authentication.1c 177 . • network – enables authorization for all network related services like: PPP. other than “default”. • reverse-access – controls access to all reverse access connections like reverse Telnet.

This would include information like system reboots. number of packets and bytes. network. • connection accounting – Provides information about all outbound connections made from the network access server. Cisco IOS does not support command acccounting with RADIUS. security analysis. SLIP. command. connection. Accounting records will include the command. • command accounting – This applies to commands which are entered in an EXEC shell. The list will define the list of accounting methods for the indicated accounting type. rlogin. accounting being turned on or off. Routers send their accounting records to the security server for storage. and reporting. Accounting AAA accounting is used for logging and tracking the activities of users (people or other network components) using a network resource. but RADIUS does not support command or system. This option will apply accounting to all commands issued at the specified privilege level. resource usage tracking. and the command that was executed. • EXEC accounting – Provides information about user EXEC sessions on the router. the usage start and stop times. etc.2 shows an example of configuring AAA authorization. There are several types of accounting which can be enabled and configured separately: exec.1c . start and stop times. As with authentication and authorization. Information in an accounting record includes the user’s identity. AAA accounting can only use the TACACS+ or RADIUS security servers for record logging. if a default list is not defined and a named list is not applied to the line then no accounting will occur for that type on that line. These logs can be used for network management. system. and the user. The information includes the username. Account records are generated based upon the level of the command not the level of the user. date. • network accounting – Provides information for PPP. If accounting is turned on for level 15 and user logged in at enable level 15 runs a level 1 exec command no accounting event will be generated. refer to the “Configuring Authorization” chapter in the IOS Security Configuration Guide [1]. IP address of access server.6. If the list was a named list then it must be applied to the appropriate lines and interfaces.Router Security Configuration Guide Section 4. date. For more detailed information about configuring authorization using AAA. and telephone number the call originated from for dial-in users. For an accounting type. • system – Provides information about system-level events. The information includes the number of packets and bytes. you configure AAA accounting by defining a list of accounting methods. and ARAP protocols. time. This includes telnet. 178 Version 1. All types are supported by TACACS+.

For more information about AAA accounting. In the case of accounting the auditing event will not occur. like RADIUS. Methods requiring only a password (e. If the Cisco IOS software is unable to complete the task due to failure to communicate with the security server or mechanism then the Cisco IOS will try the next method in the list. Cisco’s implementation of RADIUS does not support system accounting. if the user supplies a username that does not exist in the local user database then the next method on the list will be attempted.g. TACACS+. will deny access if the username does not exist. Note that system accounting will only use the default list. It is important to choose the right order for the methods on a method list.1c 179 . except for wait-start accounting which will also deny the user access for the service. A negative response from a security server will also deny access in the case of authentication and authorization and the next method in the list will not be attempted. including RADIUS and TACACS+ attributes. especially for authentication. authorization. Method lists also specify the sequence in which the security mechanisms should be used. and Kerberos security servers. Method Lists and Server Groups Method lists are used to specify one or more security protocols or mechanisms for AAA. (Note: the local user database is treated as a special case. or accounting as appropriate.Advanced Security Services etc. In the case of authentication and authorization the user will be denied access. Other mechanisms. and that a security server is specified for each accounting type which is desired. see the IOS Security Configuration Guide [1]. the first method on the list determines whether the user will be prompted for a username. This also means that a named method list will have no effect on a interface or line unless it has been applied to it. For AAA login authentication. Each accounting record is comprised of accounting AV pairs and is stored on the access control server.2 gives an example of configuring accounting. security servers are defined. the “line” method) should never be placed ahead of methods requiring a both username and password. Named method lists can then be defined and then applied to particular ports or lines to override the default behavior. If the list is exhausted then the operation will fail. Accounting can also be configured such that a user requested action can not occur until an acknowledgement is received from the security server stating that the accounting record has been saved. because the user will never be prompted for a username and the mechanism will always fail. This continues until there is a successful communication with a listed method or the list is exhausted. AAA accounting requires that AAA is enabled. For AAA the Cisco IOS software will use the first method listed to perform the authentication. When a method list is specified using the default keyword the list will be applied to all the appropriate interfaces and lines automatically.6.) Method lists can be given a specific name or can use the keyword “default”. Version 1. These lists can be used to provide backup mechanisms for when the primary security method is unavailable. Section 4.

Authentication The AAA authentication commands can be grouped into two areas which correspond to how they are applied. In order to use Cisco's AAA mechanisms you must first enable AAA services. the syntax for method lists changed slightly with the introduction of security server groups. this section focuses on using AAA for the security of remote administration. Section 4.1 and later aaa authentication login remoteauth krb5 group radius local aaa authorization exec default group tacacs+ group radius aaa accounting network default start-stop group tacacs+ The definition and use of server groups is mostly outside the scope of this guide. Router Access Control The previous section introduced authentication. for named and default method lists. there is directly controlling authentication to the router and then there are commands for providing information about the authentication process. This section will cover details of configuring AAA for controlling access to the router.6. in IOS 11.1c .0 syntax. simply insert the keyword “group” in front of the words “radius” or “tacacs+” when attempting to apply any of the examples. For detailed information about server groups.0 aaa authentication login remoteauth krb5 radius local aaa authorization exec default tacacs+ radius aaa accounting network default start-stop tacacs+ radius In IOS 12.3 briefly covers a dial-in user example.1 or later. as shown below. 4. ! AAA method list syntax for IOS 12.0. there is a small example in the next sub-section. and accounting mechanisms and how method lists are used to define the security protocol to use for a service. if you are using IOS 12.1 Security Configuration Guide.6. the command for doing this is: north(config)# aaa new-model The remainder of this section will deal with configuring the three AAA services by giving concrete examples (see Figure 4-10 on page 186) and describing the rationale behind the configuration. AAA configuration is a broad subject. The rest of this section uses the IOS 12. and the FreeRadius server was used for additional RADIUS testing. Use of RADIUS or TACACS+ requires the keyword group.4 describes security server protocols in more detail.1 and later. ! AAA method list syntax for IOS 12.2. The name of the first list is “remoteauth”. Cisco's ACS Version 2. First. consult the IOS 12. authorization. Section 4.Router Security Configuration Guide The following example shows the syntax.3 was used for testing RADIUS and TACACS+ security servers.3 and 12. The four authentication commands used for controlling access to a router are: 180 Version 1.6.

it is better to stick to generic responses and allow the administrator to look in the audit records for debugging purposes.1c 181 . This command will also require that all authentication requests to the router include a username as well as a password. Version 1. • aaa authentication password-prompt text-string changes the password prompt from “Password” to the supplied value of text-string. remember to define at least one local user (see Section 4. Another bad example would be using an informational banner to identify the router as your border router and list the protocols it accepts.1. • aaa authentication local-override is used to override all authentication method lists to look at the local database first. to guarantee all lines are protected. The authentication commands used for defining messages are: • aaa authentication username-prompt text-string changes the username prompt from "Username" to the defined value of text-string. Four authentication commands are used for giving messages to the user. There is an important point to remember when setting prompts and messages: do not give away too much information! For example. administrators will still be able to gain remote access by using a username and password defined locally on the router. • aaa authentication fail-message delimiter string delimiter defines a message to be printed when authentication fails. In a simple situation only one authentication list is required. when specifying why an authentication operation failed with the aaa authentication fail-message command. (Use with care. • aaa authentication enable default method-list can be used to control access to enable mode with the authentication mechanism.1. Including a local method will guarantee that if the security server(s) is not available. This section will concentrate on the four authentication commands for controlling access to the router. There is never really a need to use the "default" option but it could be used to be more explicit. and avoid possible default behavior changes in the IOS. If you use this approach. This list should be the default list.Advanced Security Services • aaa authentication login {default | list-name} method-list is used to specify login authentication method lists.5).) • (line): login authentication {default | list-name} is required to apply a named login authentication method list to a line. The commands deal with prompts and informational messages.4. You may choose to include ‘local’ on your method list. Using these commands in your environment may be a useful thing to do. For setting a banner on all terminals use the banner motd command as suggested earlier in Section 4. • aaa authentication banner delimiter string delimiter replaces any before system login banners with the value of string.

and should use the local user database. then any additional method which requires a username will automatically fail. first create the default list. When multiple lists are used it would be a good idea if the default list only used the local method and then named lists can be used to override the default list as appropriate. decide whether to use usernames and passwords (preferred) or to use just a password (highly discouraged). Do not use the aaa authentication enable default command since the security server pass phrase is stored in the clear and the enable secret is well protected. The following is a recommended configuration for using a RADIUS security server and the local user database as described above. should apply to the console line only. When designing your method lists. The default list should be for the limited set of administrators. then by default.1c . Accounting records can still be sent to the security server but the security server's authorization capabilities can not be used since no authentication records will be sent to the security server. To avoid being locked out of your router. make sure you add an administrator account to the local username name database before enabling AAA. For the named method list which will primarily use the security server. 182 Version 1. including VTY lines. Central(config)# username annadmin password 0 G%oD9pa$8 Central(config)# username joeadmin password 0 3MiaB-JKJ Central(config)# aaa authentication login default local Central(config)# aaa authentication login remotelist radius local Central(config)# line vty 0 4 Central(config-line)# login authentication remotelist Central(config-line)# exit Central(config)# line aux 0 Central(config-line)# login authentication remotelist Central(config-line)# exit Central(config)# In general the default list should be the most restrictive authorization list. to allow additional administrators remote access to the router. In a more complex scenario where a more limited set of administrators have access to the console line. authorization should be used to control the larger set of administrators. For accounting purposes you should use the methods which allow for usernames and assign each administrator a distinct username. The default method list designates RADIUS Central(config)# username joeadmin password 0 G0oD9pa$8 Central(config)# aaa authentication login default radius local One note about method lists for aaa authentication: whatever method is first in the list controls whether the authentication procedure will prompt for a username or not. Use the enable secret password to protect all higher privilege levels. If the first method in the list is line or enable. The second list should be a named method list and should be applied to the appropriate lines. authentication will use the local database on all lines. Important: when AAA is turned on.Router Security Configuration Guide Here is an example of setting up local username and password and AAA default login authentication parameters.

the other two (network and reverse-access) primarily deal with dial-in and dial-out access control and apply to interfaces. In scenario 2.3 on Dial-In Users provides an overview of network and reverse-access authorization. and will not be covered. authorization is useful for controlling who can access network services. • aaa authorization config-commands tells the router to do authorization on all configuration commands (this is the default mode set by the aaa authorization commands level command). AAA authorization is currently of limited use for controlling access to routers beyond the standard authentication mechanisms. • (line): authorization {arap | commands level | exec | reverse-access} {default | list-name} applies a specific authorization type to a line (note: arap is part of the network authorization type). Scenario 1 – Router with dial-in users. is also applied to lines. This section will concentrate on exec and command authorization. exec and command deal with router access control and apply to lines.6. Second. authorization can control different administrators who have access to different privilege levels on the router. First. There are two primary scenarios where authorization is useful. authorization configuration for controlling access to the router: Central(config)# aaa authorization exec default radius Central(config)# aaa authorization network default radius Scenario 2 – Router with two levels of users (exec and privileged exec) Central(config)# aaa authorization exec default radius Central(config)# aaa authorization commands 15 default radius In both scenarios there was no need to apply the authorization method lists to lines because they are using the default lists. and Section 4. Of the four authorization types. if the router is used for dial in access. arap. The router Version 1. For scenario 1 there would be additional considerations as described in the Dial-In Users section. exec is used to control all access to exec shells on the router and commands 15 is used to control access to privilege level 15 for a more restrictive set of administrators.Advanced Security Services Authorization The commands used for AAA authorization are: • aaa authorization {network | exec | commands level | reverse-access} {default | list-name} method-list turns on AAA authorization for the specified type and designates the order in which authorization methods will be applied. The no form of this command will turn off authorization on configuration commands in the EXEC mode.1c 183 . Another network type. and who can access and configure the router. etc.

The local database is populated using the username command. For a list of supported TACACS+ A-V pairs. Important: do not use the username name privilege level command since the password will be weakly protected. But there are no useful parameters to set for access to the router from lines (an exception would be for dial-in access). which are stored in a database on the security server. Even if a named method list is created and applied to the console line authorization will be ignored. For a list of supported RADIUS attributes. There is one very important exception to this.1c . Periodic generates more accounting records than newinfo since it will also include interim reports on actions in progress. (NULL usernames can occur because of accounting records on a protocol translation) • aaa accounting update {newinfo | periodic number} will allow administrators to specify when accounting records are sent to security servers. RADIUS and TACACS+ authorization both define specific rights for users by processing attributes. associated with the user. RADIUS and TACACS+. refer to the "TACACS+ Attribute-Value Pairs" appendix of [1]. and sent to the network access server where they are applied to the user's connection. • (line): accounting {arap | commands level | connection | exec} [default | list-name] can be used to apply different accounting services and levels to different lines. Accounting The commands used for AAA accounting are: • aaa accounting {system | network | exec | connection | commands level} {default | list-name} {start-stop | waitstart | stop-only | none} method-list turns on AAA's accounting services for the specified accounting type. Also. For both.1). AAA authorization does not apply to the console line. • aaa accounting suppress null-username command prevents accounting records from being generated for those users who do not have usernames associated with them. attributes are defined on the security server. Protect higher levels on the router using the enable secret command (see Section 4.Router Security Configuration Guide commands turn on the checks to query the security server on the router but the actual user to authorization privilege mapping occurs on the security server. • show accounting {system | network | exec | commands level} {start-stop | wait-start | stop-only} tacacs+ command can 184 Version 1. in the examples above if the RADIUS security server is not available no one will be able to get an exec shell and in scenario 2 no one will be able to run privilege level 15 commands. refer to the "RADIUS Attributes" appendix of [1].

• stop-only sends an accounting record at the end user process which is of an accountable type. • none specifies that no accounting records will be generated for a particular accounting type. There are two basic scenarios for accounting depending upon which security server is in use. If the start record is sent and acknowledged the user process can continue and at the end a stop accounting record will also be sent. then you must use TACACS+ accounting. and will actually be terminated. In this case the user process can not continue. Configuration of TACACS+ accounting: Central(config)# aaa accounting system default start-stop tacacs+ Central(config)# aaa accounting exec default start-stop tacacs+ Central(config)# aaa accounting exec remoteacc wait-start tacacs+ Central(config)# aaa accounting commands 15 cmdacc stop-only tacacs+ Central(config)# aaa accounting connection default start-stop tacacs+ Central(config)# line vty 0 4 Central(config-line)# accounting exec remoteacc Central(config-line)# accounting commands 15 cmdacc Central(config)# line aux 0 Central(config-line)# accounting exec remoteacc Central(config-line)# accounting commands 15 cmdacc Version 1. This is all done in the background and the user process will continue regardless of the outcome of the accounting attempt. AAA allows for four levels of accounting as set by the aaa accounting command: • start-stop accounting sends records when the accounting type starts and stops.Advanced Security Services be used to show active connection information. This is not a configuration command but is worth mention. As mentioned earlier Cisco's RADIUS implementation does not support system and command accounting. • wait-start accounting sends an accounting record at the start and stop of each specified type. If your security policy calls for keeping a record of every router command. Do not use wait-start in any accounting method list intended for the console line! A basic recommendation would be to use wait-start for remote users and start-stop for local users. For command accounting stop-only will provide the necessary coverage and will greatly reduce the number of accounting records. Important: if wait-start accounting is specified on an interface or line and no security server is available for receiving the accounting record then the user process using that interface or line will be locked out. if the start accounting record can not be recorded.1c 185 .

Router Security Configuration Guide

Configuration of RADIUS accounting:
Central(config)# aaa accounting exec default start-stop radius Central(config)# aaa accounting exec remoteacc wait-start radius Central(config)# aaa accounting connection default start-stop radius Central(config)# line vty 0 4 Central(config-line)# accounting exec remoteacc Central(config)# line aux 0 Central(config-line)# accounting exec remoteacc

Since remote administration is more dangerous than console administration, the configurations above add extra accounting to the remote lines. Part of the extra protection is requiring that before a remote user can get an exec shell an audit record must be recorded into the security server. Note: the aux line configuration is not required if the aux line is disabled as suggested in Section 4.6.2. Also, for information about RADIUS Attributes and TACACS+ AV Pairs for use in accounting, refer to the appendices in the Cisco Security Configuration Guide [1].

Putting It Together
This section will put together the AAA mechanisms from earlier in this section and will apply them to the configuration of the Central and South Routers. The Central router is between the facility backbone and the specific part of the infrastructure. The South router acts as the first layer of defense to a well protected enclave.

Facility Network

eth 0


eth 1

Authentication Server

eth 0/0

eth 0/1

eth 0/0


eth 0/1

Protected Enclave

Figure 4-12: Routers and their Authentication Server

Authorization will not be used in these examples since all the administrators in these examples need configuration access and there is no dial-in access. For a more


Version 1.1c

Advanced Security Services

complete example, including authorization and some discussion of dial-in security concerns, see Section 4.6.3. Central Router Configuration (IOS 12.0):
Central(config)# enable secret 3rRsd$y Central(config)# username fredadmin password d$oyTld1 Central(config)# username bethadmin password hs0o3TaG Central(config)# username johnadmin password an0!h3r( Central(config)# service password-encryption Central(config)# banner motd ^T Legal Notice: Access to this device is restricted. . . ^T Central(config)# radius-server host Central(config)# radius-server key i*Ma5in@u9p#s5wD Central(config)# aaa new-model Central(config)# aaa authentication login default radius local Central(config)# aaa accounting exec default start-stop radius Central(config)# aaa accounting exec remoteacc wait-start radius Central(config)# aaa accounting connection default start-stop radius Central(config)# access-list 91 permit log Central(config)# access-list 91 deny any log Central(config)# line con 0 Central(config-line)# transport input none Central(config-line)# exec-timeout 5 0 Central(config-line)# login local Central(config-line)# exit Central(config)# line vty 0 4 Central(config-line)# access-class 91 Central(config-line)# exec-timeout 5 0 Central(config-line)# login local Central(config-line)# transport input telnet Central(config-line)# accounting exec remoteacc Central(config-line)# exit Central(config)# line aux 0 Central(config-line)# transport input none Central(config-line)# login local Central(config-line)# exec-timeout 0 1 Central(config-line)# no exec Central(config-line)# end

The first thing to do when configuring access to a router is to setup the local access. The enable secret command sets the password on the privileged exec level and the username commands setup all the local accounts. Now when AAA is turned on the default authorization will not lock out the console. The message of the day should be used to provide the legal document for controlling access to the device and allowing for monitoring. This message should be generic and hopefully the same on all of your routers, firewalls, servers, workstations, etc.

Version 1.1c


Router Security Configuration Guide

Next configure the security server and turn on AAA mechanisms. Since the shared secret to the RADIUS server is stored in the clear do not use the same shared secret for the router with any other device. Since communications to the security server are protected and the connection does not go outside the corporate boundary it is acceptable to allow communications to the server outside the router. With the aaa authentication login command make sure local is in the list as described earlier. Also, notice that the default accounting for exec is set to start-stop and that a named list was created for wait-start. This way by applying the named list to external connections and allowing the default list to automatically apply to console you will not be locked out of the router. Use connection accounting to track outbound connections generated by users logged onto the router, these should be minimal. Create and apply an access-list to the VTYs to limit remote access to internal networks only and if possible limit the remote hosts by actual host IP addresses instead of a network address. Issue the login local command on the console and vtys in case AAA services get turned off. This will continue to allow limited remote access based upon the local database and will be ignored while AAA mechanisms are still running. Also limit remote access to telnet only and limit the connection idle time to 5 minutes. The auxiliary port is disabled in this example. If a TACACS+ server was used in this example instead of the RADIUS server then system accounting would have also been specified. Command level accounting could have been applied as well but would probably not be needed here. South Router Configuration:
South(config)# enable secret rI^3r6Ed South(config)# username bethadmin password hs0o3TaG South(config)# username johnadmin password an0!h3r( South(config)# banner motd ^T . . ^T South(config)# tacacs-server host South(config)# tacacs-server key Ir3@1yh8n#w9@swD South(config)# aaa new-model South(config)# aaa authentication login default tacacs+ local South(config)# aaa accounting exec default start-stop tacacs+ South(config)# aaa accounting exec remoteacc wait-start tacacs+ South(config)# aaa accounting connection default start-stop tacacs+ South(config)# aaa accounting system default start-stop tacacs+ South(config)# aaa accounting commands 15 default stop-only tacacs+ South(config)# access-list 91 permit log South(config)# access-list 91 permit log South(config)# access-list 91 deny any log South(config)# line con 0 South(config-line)# transport input none


Version 1.1c

Advanced Security Services

South(config-line)# South(config-line)# South(config-line)# South(config)# line South(config-line)# South(config-line)# South(config-line)# South(config-line)# South(config-line)# South(config-line)# South(config-line)# South(config)# line South(config-line)# South(config-line)# South(config-line)# South(config-line)# South(config-line)#

exec-timeout 5 0 login local exit vty 0 4 access-class 91 exec-timeout 5 0 login local transport input telnet login authentication remotelist accounting exec remoteacc exit aux 0 transport input none login local exec-timeout 0 1 no exec end

As in the first example start by setting up local access to the router. The enable secret command sets the password on the privileged exec level and the username commands setup all the local accounts. In this case there may be fewer local accounts since this router is the first lines of defense to a secure enclave. Again, when AAA is turned on the default authorization will not lock out the console. The Message of the Day should be used to provide the legal document for controlling access to the device and allowing for monitoring. This message should be generic and hopefully the same on all of your routers, firewalls, servers, workstations, etc. Next configure the security server and turn on AAA mechanisms. Since the shared secret to the TACACS+ server is stored in the clear do not use the same shared secret for the router with any other device. Since communications to the security server are protected and the connection does not go outside the corporate boundary it is acceptable to allow communications to the server outside the router. With the aaa authentication login command make sure local is in the list as described earlier. Notice that the default accounting for exec is set to start-stop and that a named list was created for wait-start. This way by applying the named list to external connections and allowing the default list to automatically apply to console you will not be locked out of the router. Use connection accounting to track outbound connections generated by users logged onto the router, these should be minimal. Also, include system and commands 15 accounting since this router is providing protection to a special enclave. As before, create and apply an access-list to the vtys to limit remote access to internal networks only and if possible limit the remote hosts by actual host IP addresses instead of a network address. Issue the login local command on the console and vtys in case AAA services get turned off. This will continue to allow limited remote access based upon the local database and will be ignored while AAA mechanisms are still running. Also limit remote access to telnet only and limit the connection idle time to 5 minutes. The auxiliary port is disabled in this example.

Version 1.1c


Router Security Configuration Guide

If a RADIUS server was used in this example instead of the TACACS+ server then system and command accounting would not be specified.

4.6.3. Dial-In Users
AAA services were designed with remote network access in mind. This includes remote access to routers as well as to network services like PPP. AAA using RADIUS is one of the primary means by which this is accomplished by Internet Service Providers (ISP's). Controlling access for dial-in users is similar to controlling access to the router but there are different protocols that are used. Additionally, although it is not shown, it is highly recommended that when dial-in access to the network or router is in use, that AAA services should be used in conjunction with a one-time password or similar token technology. Some important commands for controlling dial-in users are: • aaa authentication ppp {default | list-name} <method-list> is used to specify PPP authentication method lists. • aaa authorization {network | exec | commands level | reverseaccess} {default | list-name} <method-list> turns on AAA authorization for the specified type and designates the order in which authorization methods will be applied. In this case we are particularly interested in turning on network authorization. • aaa accounting {system | network | exec | connection |
commands level } {default | list-name} {start-stop | waitstart | stop-only | none} method-list turns on AAA's accounting

services for the specified accounting type. For dial-in users network needs to be used. • aaa processes number command is used to specify the number of background processes to start to handle concurrent authentication and authorization requests. • (interface): ppp authentication {pap | chap | pap chap | chap
pap} [if-needed] {default | list-name} [call-in] [one-tone]

command is used to enable pap, chap, or both forms of authentication on the selected interface. • (interface): ppp authorization {default | list-name} command is used to apply a ppp authorization list to the selected interface. • (interface): ppp accounting [default | list-name] command is used to apply accounting methods to the PPP service on the selected interface. The example below gives one potential application of AAA services for dealing with dial-in services (Note: this example is not complete). Figure 4-13 shows the relevant portion of the network, and the configuration for East is shown after it.


Version 1.1c

Advanced Security Services

Facility Network

eth 0

net access

eth 1

Authentication Server

eth 0/0


User Host

eth 0/1

Telephone Network



Remote Host

Figure 4-13: Router East in the Network
East(config)# enable secret t!tRd-1rZZ East(config)# username fredadmin password d$oyTld1 East(config)# username bethadmin password hs0o3TaG East(config)# banner motd ^T LEGAL NOTICE: Use of this device restricted to authorized persons. This device is subject to monitoring at all times, use of this device constitutes consent to monitoring. ^T East(config)# radius-server host East(config)# radius-server key i3dRc8sRv(@oeU4) East(config)# aaa new-model East(config)# aaa authentication login default radius local East(config)# aaa authorization exec default radius East(config)# aaa authorization network default radius East(config)# aaa accounting exec default start-stop radius East(config)# aaa accounting exec remoteacc wait-start radius East(config)# aaa accounting connection default start-stop radius East(config)# aaa accounting network default wait-start radius East(config)# access-list 91 permit log East(config)# access-list 91 permit log East(config)# access-list 91 deny any log East(config)# line con 0 East(config-line)# transport input none East(config-line)# exec-timeout 5 0 East(config-line)# login local East(config-line)# exit East(config)# line vty 0 4 East(config-line)# access-class 91 East(config-line)# exec-timeout 5 0 East(config-line)# login local East(config-line)# transport input telnet East(config-line)# accounting exec remoteacc

Version 1.1c


Security servers are important to Cisco network gear when centralized administration is required or when authorization and accounting services are needed. 4. In addition. AAA is the mechanism used to establish communications with security servers.2) to carry remote administrator access to the router (which is required to protect the username and password traveling across the facility backbone in the clear). TACACS+. RADIUS Remote Authentication Dial In User Service (RADIUS) is an IETF proposed standard (RFC 2865) for securing network components. authorization.1. and Kerberos.6. The asynchronous interface contains the commands necessary for configuring AAA authentication for the ppp protocol. However Cisco's RADIUS 192 Version 1. particularly the “Dial Solutions Configuration Guide”. RADIUS was designed with Dial In access control in mind and the accounting features are very flexible along these lines. Configuring dial-in services is far too complex a subject to be dealt with in depth in this guide. Also the AAA authorization and accounting default commands for network will also apply to the ppp traffic as it traverses the line. RADIUS protects the communications using a shared secret. If a TACACS+ server was used in this example instead of the RADIUS server then system accounting would have also been specified.4. The following descriptions will only discuss items which are different from the Putting It Together examples in the previous section. RADIUS is a distributed client/server based architecture used to pass security information between access points and a centralized server. 2) the terminal server lines have not been configured (and will need to have the remoteacc accounting list applied) and. Security Server Protocols In Cisco routers and network access servers. AAA authorization for exec and network was added to separate the privileges for network users and router administrators. 3) the asynchronous interface configuration needs completed (if the aux port is not used as an asynchronous interface disable it see Section 4.Router Security Configuration Guide East(config-line)# exit East(config)# interface async 1 East(config-if)# encapsulation ppp East(config-if)# ppp authentication chap East(config-if)# end In this example there are several items left incomplete: 1) the IPSec tunnel to Central has not been configured (see Section 5.1c . Consult the Cisco IOS documentation. Cisco supported security servers are RADIUS. Command level accounting could have been applied as well but would probably not be needed here. This section only provides one example for a possible network access server configuration. accounting was added for recording network events.4). RADIUS can be used to provide authentication. for more details. and accounting services.

For a complete list of RADIUS router configuration commands see the “RADIUS Commands” section in [1].18 auth-port 1812 Central(config-sg)# server 14. Some RADIUS servers use the old standard port 1645 for authentication.2.1c 193 . East(config)# radius-server host 14.2.18 key i*Ma5in@u9p#s5wD Central(config-sg)# end Central# Version 1.6.Advanced Security Services client does not support auditing of command or system events on the router or network access server.1. while others use the new standard port of 1812. Use the auth-port parameter to cause IOS to send RADIUS requests to the server on that port.1. • radius-server key string sets the RADIUS server shared encryption key.a group with one server in it Central(config)# aaa server group radius radGroup1 Central(config-sg)# server 14. Central(config)# ip radius source-interface loopback0 Central(config)# radius-server host 14. To perform authentication and authorization a RADIUS server uses attributes.2.2. For more details see the Security Configuration Guide on "Configuring RADIUS" and "RADIUS Attributes" sections for more details. The shared secret key should be at least 16 characters long and follow the other rules for a good password as described in Section 4. These groups may be useful for large enterprises. so make it longer than a typical password.4. users never have to type this string. ! RADIUS example . use the command aaa server group. The example below shows how to set up RADIUS on the router Central. As a minimum when setting up a RADIUS server on a Cisco device the host address and shared secret must be configured as well as turning on and configuring AAA on the device. This is accomplished using the commands listed: • radius-server host {hostname | ip-address} [auth-port port-number] [acct-port port-number] command specifies the radius server's hostname or IP address and the ports to use for authentication (authorization) and accounting. or later.18 auth-port 1812 Under IOS 12.18 Central(config)# radius-server key W@t7a8y-2m@K3aKy RADIUS servers are freely available and are in extensive use. as shown below. if you have defined it as described in Section 4. Also. To define a server group. you can define named groups of RADIUS servers. The secrecy and quality of this key is critical to the security of your RADIUS installation. IOS always uses 1645 unless you specify otherwise. the RADIUS service should be bound to the loopback0 interface.6. These attributes can be configured to allow/deny access to various router and network services. where different sets of security servers are used for different groups of users or different purposes.

4.1.2. TACACS+ is implemented by Cisco using the AAA mechanisms and provides for the centralized validation of users using routers and network services. TACACS+ was designed with Cisco implementations in mind so it offers a wide range of AAA services including full auditing of Cisco AAA accounting events. The [port integer] can be used to specify a new port number. • tacacs-server key string command sets the default TACACS+ shared encryption key. TACACS+ uses attribute-value pairs for controlling authentication and authorization services. so make it longer than a typical login password.Router Security Configuration Guide To use a server group. you can define named groups of TACACS+ servers. name it in a method list instead of the default group ‘radius’. and users never have to type it. Extended TACACS. These groups may be useful for large enterprises.6. where different sets of security servers are 194 Version 1. where the TACACS+ server is running. The security of TACACS+ depends on this secret. For a complete list of TACACS+ router configuration commands see the "TACACS. Simple example for Central: Central(config)# tacacs-server host 14. and TACACS+ Commands" section in the "Security Command Reference". For more details on the TACACS+ and attribute-value pairs see the Security Configuration Guide sections “Configuring TACACS+” and “TACACS+ Attribute-Value Pairs”. TACACS+ protects communications using a shared secret key between the network device and central server.18 Central(config)# tacacs-server key W@t7a8y-2m@K3aKy TACACS+ implementations are available through Cisco Secure ACS and Cisco also offers a free implementation as well. Central(config)# aaa authentication login VTlogin group radGroup1 TACACS+ Terminal Access Controller Access Control System plus (TACACS+) is the most recent Cisco security protocol designed to provide accounting and flexible control of authentication and authorization services. The shared secret key should be at least 16 characters long and follow all the rules for a good password as described in Section 4.1 or later. These attribute-value pairs are configured on the server and used by the router authorization mechanism to control access to network services. The primary commands used for configuring TACACS+ on a Cisco router are: • tacacs-server host {hostname | ip-address} [port portnumber] [key string] command can be used to specify the host. IP address or DNS name. The key string parmeter sets the secret key for this TACACS+ server host overriding the default but should follow same creation rules as the default. Under IOS 12.1c .

but more details can be found in the IOS documentation [1]. Some modern operating systems come with Kerberos built in. Before attempting any of the step below.2. Kerberos infrastructures are already in wide use. To define a server group. consult the IOS 12. if your Telnet client supports Kerberos encryption. Kerberos provides strong authentication for client/server applications by using secret-key cryptography.2. The communications between the router and the Kerberos security server are encrypted. ! TACACS+ Example . If you already have a Kerberos infrastructure in place.6.1c 195 . Central(config)# aaa authentication login VTlogin group myTacGroup Kerberos Kerberos was developed by the Massachusetts Institute of Technology (MIT) and is standardized by the IETF as a network authentication system in RFC 1510.10. Configuration of Kerberos installations based on MIT Kerberos are already explained in the Cisco IOS documentation. For a good introduction to these topics. Host configuration for using MIT Kerberos is not covered in this guide.e. Kerberos can also be used to perform EXEC shell authorization using Kerberos Instance Mapping. person or network component) on unprotected networks. These two topics are outside the scope of the Kerberos coverage in this guide. This mechanism can verify the identities of two users (i.18 key Gx98-vAR1bv*u Central(config-sg)# server 14. then this form of centralized authentication may be a way to gain excellent security for remote administration. as shown below. consult [8].Advanced Security Services used for different groups of users or different purposes. This authentication is performed using a trusted third-party service using conventional (secret key) cryptography. There are free open sources versions of Kerberos available as well as commercially supported products.1 Security Configuration Guide for more information.39 key t777+08cdcoWW Central(config-sg)# end Central# When you want to include the servers of a particular group in a method list. use the command aaa server group. This section assumes basic familiarity with Kerberos administration and security concepts.a group with two servers in it Central(config)# aaa server group tacacs+ myTacGroup Central(config-sg)# server 14. After the two parties have been authenticated (in this case. simply use the group name instead of the default name ‘radius’ or ‘tacacs+’. Configuration of a Microsoft Windows 2000 Server acting as the Kerberos authentication server is covered below. Kerberos can provide very effective confidentiality and data integrity services. In this system a client would request the credentials of the party they wish to contact from the trusted authentication service. Note that Kerberos only allows for limited authorization capabilities and no accounting. as well as in RFC 1510 [5] and in Tung’s book [8]. the router and the adminstrator). make sure that the IOS installed on your router supports Version 1.

196 Version 1. To make it work with a Cisco router. and select “New”. perform the following steps on your Windows server: 1. Update or confirm the DNS entries for the KDC and the router.Router Security Configuration Guide Kerberos. Install the Kerberos support tools from the Windows 2000 installation media. right click on the “Users” folder.exe”. not for any user.1c . 3. If necessary. (For example. create the user accounts on the server for administrators that will access the router. in global config mode type the word kerberos and then type a question mark. Create a user account for the router.) A Windows 2000 Server configured to be a Domain Controller automatically has the Kerberos Key Distribution Center services installed and running on it. if you get several choices then your IOS supports Kerberos. then “user”. Open up the “Active Directory Users and Computers” tool located in the “Control Panel\Administrative Tools” folder. (Note: this is a Kerberos identity for the router.) 4. The tools are found in “support\tools\setup. 2.

Advanced Security Services

5. Check the Kerberos settings for logins; use the settings shown below in the column “Effective Setting”. For more information, consult the NSA Guide to Windows 2000 Kerberos Settings [6].

6. Use the Windows ktpass command, installed in step 1, to create the host’s keytab file, map the router to its account, and set its password.

7. Install the keytab file on the router. This must be done using the IOS kerberos srvtab command, as shown below. Once you are sure that your router supports Kerberos, follow the steps listed below in global config mode. 1. Define the Kerberos realm.
kerberos local-realm kerberos-realm

2. Designate the Kerberos KDC to use in the realm, along with port number.
kerberos server kerberos-realm {hostname | ip-address} [port-number]

3. Map an optional host name or DNS domains to the realm.

Version 1.1c


Router Security Configuration Guide

kerberos realm {dns-domain | host} kerberos-realm

4. Define the preauthentication method.
kerberos preauth authentication-method

The recommended method is encrypted-kerberos-timestamp. 5. Generate a local private DES key. The key-password should be 8 randomly-chosen characters.
key config-key 1 key-password

The key will be used to encrypt the Kerberos secret key in the router’s stored configuration. (This key is stored in the router’s NVRAM, but cannot be recovered or extracted.) 6. Load the keytab file from a server, link channel, or local file.
kerberos srvtab remote { URL | host filename }

This command supports a wide variety of means for downloading the srvtab file, including TFTP, FTP, and more. TFTP is the default. 7. Create a login authentication model, specifying Kerberos as the mechanism to use first.
aaa authentication login {default | list-name} krb5 [ {other-mechanisms} ]

The two examples below show two different ways of conveying the Kerberos keytab file, generated by the Windows 2000 ktpass command, over to the router. Neither approach is perfect -- the ideal approach would be to load the keytab file directly over the serial link, but unfortunately that is not supported in any version of IOS that the authors of this guide have been able to test. Kerberos Example 1: Network Download The transcript below shows an example of following the Kerberos setup procedure on router Central. In this case, the Kerberos KDC provides service on IP address, and a TFTP server is on IP address (Note: Windows 2000 installations typically do not include TFTP servers. You will need to use a commercial TFTP server, or distribute the file from a Unix system equipped with a TFTP server.)
Central# config t Enter configuration commands, one per line. End with CNTL/Z. Central(config)# kerberos local-realm KERBEROS.NSA.GOV Central(config)# kerberos server KERBEROS.NSA.GOV Central(config)# kerberos realm KERBEROS.NSA.GOV Central(config)# kerberos realm KERBEROS.NSA.GOV Central(config)# kerberos preauth encrypted-kerberos-timestamp Central(config)# key config-key 1 aW.-8(xZ Central(config)# kerberos srvtab remote Central.keytab Loading Central.keytab from (via Ethernet0/1): ! [OK - 78/4096 bytes] Central(config)#


Version 1.1c

Advanced Security Services

The content of the keytab file is very sensitive, because it contains the long-term Kerberos secret key that the router will use to communicate with the KDC. In general, it is not a good idea to transfer the keytab file from your server to the router over TFTP (or FTP, or any other plaintext network protocol). Unless this part of your configuration takes place on an isolated lab or management network, do not use this method to distribute the keytab file. Instead, use (1) the console serial download method described below, or (2) the SCP protocol instead of TFTP (this requires IOS support for SCP, and it requires setting up SSH as described in Section 5.5). Kerberos Example 2: Console Link Download This example shows how to convey the keytab file securely from your server to the router over the console serial link using the YModem protocol.
Central# copy ymodem: flash:central.key **** WARNING **** . . Proceed? [confirm]y Destination filename [central.key]? central.key Erase flash: before copying? [confirm]n Max Retry Count [10]: 15 Perform image validation checks? [confirm]n Ymodem download using crc checksumming with NO image validation Continue? [confirm]y Ready to receive file...........C 4294967295 bytes copied in 23.692 secs (0 bytes/sec) Central# Central# config t Enter configuration commands, one per line. End with CNTL/Z Central(config)# kerberos local-realm KERBEROS.NSA.GOV Central(config)# kerberos server KERBEROS.NSA.GOV Central(config)# kerberos realm KERBEROS.NSA.GOV Central(config)# kerberos realm KERBEROS.NSA.GOV Central(config)# kerberos preauth encrypted-kerberos-timestamp Central(config)# key config-key 1 .XT9+se% Central(config)# kerberos srvtab remote flash:central.key Central(config)# exit Central# ! optional steps: wiping the keytab from flash Central# delete flash:central.key Delete filename [central.key]? central.key Delete flash:central.key? [confirm]y Central# Central# ! the squeeze command may not be supported on all routers Central# squeeze flash: Squeeze operation may take a while. Continue? [confirm]y . . Central#

Version 1.1c


Router Security Configuration Guide

Kerberos and AAA Once you have downloaded the keytab file, you can designate Kerberos as your main AAA authentication method.
Central(config)# aaa new-model Central(config)# aaa authentication login default krb5 local Central(config)# exit Central#

This section presents only a very cursory look at Kerberos authentication. For more information about administering Kerberos networks, consult [6], [7], and [8].

4.6.5. References
[1] Cisco Systems, Cisco IOS 12.0 Network Security, Cisco Press, 1999. This book provides a detailed reference for all the security features in Cisco IOS 12. It includes a great deal of information about AAA, including a section on configuring Kerberos. The same information is also available in the on-line documentation, in the Cisco IOS Security Configuration Guide. [2] Cisco System, Cisco IOS 12.0 Dial Solutions, Cisco Press, 1999. This documentation volume provides detailed information on setting up modems and dial-up networking facilities. [3] Rigney C., et. al. “Remote Authentication Dial In User Service (RADIUS)” RFC 2865, June 2000. This is the Internet RFC that defined the core RADIUS protocol. [4] Carrel, D., and Grant, L. “The TACACS+ Protocol Version 1.78”, Cisco Systems, January 1997. available at: This is the draft RFC that would have standardized the TACACS+ protocol. It explains the operation of the protocol in great detail. [5] Kohl, J., “The Kerberos Network Authentication Service (V5)”, RFC 1510, September 1993. This is the Internet RFC that defines the Kerberos authentication protocol. [6] Opitz, D. “Guide to Windows 2000 Kerberos Settings” NSA, July 2001. available under: This guide describes prudent Kerberos security settings for Windows 2000.


Version 1.1c

Advanced Security Services

[7] “Step-by-Step Guide to Kerberos 5 Interoperability,” Windows 2000 Step-byStep Guides, Microsoft Corporation, 2002. available at:

This article describes how to use Windows 2000 Kerberos with other Kerberos implementations. [8] Tung, B., Kerberos - A Network Authentication System, Addison-Wesley, 1999. This slim handbook provides a good overview of Kerberos.

Version 1.1c


Router Security Configuration Guide

4.7. Collected References
The list below describes the major references and sources of information for the material presented here in Section 4.

4.7.1. Books and Manuals
Cisco Systems, IOS 12.0 Configuration Fundamentals, Cisco Press, 1999. Basic configuration guide for IOS 12, includes good information on using the IOS command interface, basic IOS commands, and much more. Cisco Systems, Cisco IOS Network Security, Cisco Press, 1998. This book is the security configuration manual and command reference for IOS 11.3. It includes extensive coverage of access management, AAA, and related topics. Available on the Cisco Documentation CD as two documents: the “Security Configuration Guide” and the “Security Command Reference”. Cisco Systems, Cisco IOS 12.0 Network Security, Cisco Press, 1999. This book is the security configuration manual and command reference updated for IOS 12.0. It includes extensive coverage of access management, AAA, IPSec, and related topics. Available on the Cisco Documentation CD. Akin, T., Hardening Cisco Routers, O’Reilly & Associates, 2002. A pragmatic and detailed guide to securing Cisco routers; includes detailed examples. Held, G. and Hundley, K., Cisco Security Architectures, McGraw-Hill, 1999. This book includes excellent general advice about router and router-related network security, in addition to its Cisco-specific material. Held, G. and Hundley, K., Cisco Access List Field Guide, McGraw-Hill, 1999. Access lists are critical to most aspects of Cisco IOS security. This book is a detailed, practical guide to creating and understanding access lists. Innokenty, R., Cisco Routers for IP Routing: Little Black Book, Coriolis Group, 1999. This practical little book includes great advice on managing routes and routing protocols, mostly oriented toward IOS 11.2 and 11.3. Chappell, L. editor, Advanced Cisco Router Configuration, Cisco Press, 1999. Good coverage of advanced Cisco configuration issues, including extensive material on access lists and OSPF.


Version 1.1c

Cisco IOS 11. Cisco Press. B. Cisco IOS Release: The Complete Reference.html A good overview article on tightening up the security on a typical Cisco router running IOS Routing in the Internet. April This short but highly prescriptive document distills a great deal of Cisco IOS security practice into an example configuration. “Increasing Security on IP Networks” Cisco Internetworking Case Studies. including good coverage of critical security topics. and Smith. A detailed exploration of the SNMP management information base. Unbelievably detailed information on Cisco IOS release versions. 1998. “Secure IOS Template . and Perkins. Release Notes. Articles and Papers Thomas.Advanced Security Services Coulibaly. and how routing works in the Internet. Prentice-Hall.1”. features in 4.3 or later.1c 203 ..Version 3.. “Unicast Reverse Path Forwarding”. 1999... “Improving Security on Cisco Routers”.7. protocols. Huitema. Understanding SNMP MIBs. C. 1996. 1st edition. Greene. R. 2000. Cisco ISP Essentials. November 2003.htm Initial documentation on unicast reverse-path forwarding verification. Cisco Press.. E. P. Version 1. software/ios111/cc111/uni_rpf. and upgrade paths.M. 2nd Edition. M. it includes a good explanation of the concepts. A deep and detailed textbook about IP routing technologies. available at: An old but useful article on using a Cisco router to protect a network Cisco Security Advisories. It has also been published as a book. Includes some coverage of access lists and passwords. the release management process. available from Cisco Press. available at: http://www. Addison-Wesley. including both standard and vendor-specific structures. McGinnis.2. available at: http://www. Cisco Systems. available at: http://www. 2002. This detailed guide explains a great deal about operational use of Cisco routers in the Internet Service Provider environment.

and local trusted networks. 5. By focusing on this fundamental building block of IP networking. etc. the network administrator or security engineer can be overwhelmed. Routers used for supplying packet protection are almost always positioned as gateway or border devices. The vast amount of available literature and the technical jargon can cause an administrator to ignore available security features altogether. Each packet passing through. It works in conjunction with the standards-based Internet Key Exchange (IKE) protocol to provide the users a very solid IP security foundation. or created by the router has source addresses and is carrying data which may need some form of protection. we can devote our energy to providing you with some basic cryptographic concepts. Role of the Router in Inter-Network Security When considering the task of joining IP security with IP router functionality. Thus.1. Since the adoption of the IETF IP Security (IPSec) standards. These facilities offer additional security services that can contribute to the secure operation of entire networks or communities. and the specific Cisco IOS commands that implement them. servers. This standards-based approach allows for interoperability between Cisco routers and other IP security products. In 1996. These devices sit between untrusted networks.1c . To reduce this daunting task to one which is manageable and easily understandable. This proprietary solution was a stopgap effort for customers until a standards-based solution was in place.2. such as the Internet. this section of the guide will focus on the concept of “packet protection”. it did not allow Cisco products to interoperate with other IP security products. This framework has been scrutinized by many skilled evaluators in industry and academia. Cisco released IOS version 11.Router Security Configuration Guide 5. While it provided some level of packet protection for Cisco-to-Cisco communications. both Cisco (in IOS 11. e. non-Cisco routers. IPSec tunnels can be constructed between two routers’ interfaces using the IPSec protocol framework. which included the Cisco Encryption Technology (CET). 204 Version 1. These can then be easily incorporated into current router configurations to help meet specific security requirements. firewalls.3 and above) and other IP product manufacturers have implemented and offered IPSec solutions for packet protection to their customers.g. Advanced Security Services This section describes some Cisco IOS facilities that are not central to the task of securing a router.

20 host 14.20.0. the routers may be configured using several different modes of operation. In the basic mode of operation.12.1.g. Establishing the IPSec Protection Parameters Establishing a Common IKE Authentication Key Prior to establishing an IPSec tunnel between two routers.0. Building IPSec Tunnels Building IPSec tunnels between two Cisco routers will involve entering three sets of information into each router’s running configuration files. This authentication decision is made in the IPSec framework using the IKE protocol. access-list 100 permit 50 host 7.1.0. 5.0. any access list restrictions on these ports or protocols should be removed or changed to allow the IPSec packets to be transmitted and received by the participating routers. the administrator can update and/or change the current running configuration. anyone with access to the router can view selected information about the current running configuration. The security guidance of this section does not exhaustively cover all IPSec options. Version 1. In the example that follows.20 host 14.12. each router authenticates the peer IP address with which they are building a tunnel. and the Remote router. IP Network Security Prior to establishing an IPSec configuration on the router. However. the external interfaces of the North router. a pre-shared key) to do it.20.1. which algorithms to use) and the appropriate Cisco IOS commands to implement them in an easy-to-follow. For the example in this section.2. While IKE has several ways it can authenticate the two routers to each other.1.12. we will only discuss how it uses a jointly held secret value (i.20 access-list 100 permit udp host 7.e. it provides a set of options (e. step-by-step example to help you set up and test IPSec on your network.2.1c 205 .20 access-list 100 permit 51 host 7. The sets can be labeled as: 1. Establishing a common IKE Authentication Key 2.20 host 14. In the privileged EXEC mode. Establishing an IKE Security Policy 3.2. Since IPSec utilizes IP protocols 50 and 51. see Section 4. we assume the routers have two modes of operation: basic mode and privileged EXEC mode.2. The example below illustrates the ACL rule syntax for permitting incoming IPSec traffic.20 eq 500 Also. 7.1. 14.2. and the User Datagram Protocol (UDP) port 500 in its communications. for operational security we HIGHLY recommend that IKE authentication decisions be made using IPSec authentication schemes in conjunction with digital certificates. will be used to help demonstrate the concepts (see Figure 4-1).Advanced Security Services 5. For more information about command modes.12. Rather.2.1. certain network and current router configuration checks should be made to eliminate any router connectivity problems.

We strongly recommend using difficult-to-guess combinations of characters.20 Remote(config)# exit Remote# When entering new configuration information into the router it is always a good idea.0 Security Configuration Guide [2] for details on the other IKE options. the secret key is “01234abcde”.Router Security Configuration Guide Consult the Cisco IOS 12. the network administrators for the North and Remote routers (possibly the same person) should enter the key into their routers. one per line. The syntax for the crypto isakmp command is: crypto isakmp key key-value address destination-ip-address. End with CNTL/Z. These keys should be obtained out-of-band by each of the routers’ administrators.1. End with CNTL/Z. there must be at least one policy in common between them. In order for two routers to be interoperable.0. as shown below. For this example. (Note: the router used for part of this example is named “Remote”. Do not use a remote administration connection to enter sensitive IPSec parameters – use a local console connection. numbers. Once the keys are securely held. to check and see if the router has received the intended configuration information. One way to verify that the pre-shared keys were properly entered is to display the router’s running-configuration and look for the preshared key entered above. and that name appears in all the prompts. use the crypto isakmp command in global configuration mode.2.) To use pre-shared keys for making authentication decisions in IKE. Remote(config)# crypto isakmp key 01234abcde address 14. North# config t Enter configuration commands. To enter the keys. each router must possess the same secret key.20 North(config)# exit North# and Remote# config t Enter configuration commands. Establishing an IKE Security Policy Each router contains a list of IKE security polices. one per line. These policies capture information needed by the IKE protocol to help build a secure IPSec tunnel between the two routers.1c . Each necessary parameter for the policy is listed 206 Version 1. This can be done using the show running-config command in privileged EXEC mode. and punctuation symbols to build operational pre-shared keys. after entering the new information. North(config)# crypto isakmp key 01234abcde address 7.12.

pre-shared keys) • Diffie-Hellman group – used for computing the encryption key (choices: #1 (768 bit modulus). 3DES doesn’t provide the needed performance) always use 3DES. North# North# config t Enter configuration commands. to protect information between two peers over a hostile. unprotected network (e. #5 (1536 bit modulus)). one per line. the Internet). The DES algorithm is not acceptable. however.1c 207 . North(config)# crypto isakmp policy 1 ! The policy number may be an integer between 1 and 65. ! then use the following command to select DES for encryption ! North(crypto-isakmp)# encryption des North(crypto-isakmp)# hash sha North(crypto-isakmp)# authentication pre-share North(crypto-isakmp)# group 2 North(crypto-isakmp)# lifetime 86400 North(crypto-isakmp)# exit North(config)# exit North# and Version 1. Unless you have a very sound reason to use DES. e. so use 3DES for such cases. #2 (1024 bit modulus).g. and community of interest data separation is needed.g. otherwise use group #2 • security association lifetime – lifetime. that a tunnel should remain in place before it is automatically rebuilt (default: 86400 (one day)) The administrators for the North and Remote routers should enter the IKE security policy into their routers using the commands shown below. (e.g. with ! the priority given to lower numbers North(crypto-isakmp)# encryption 3des ! If the user’s version of the IOS only supports the DES ! algorithm. group #5 should be used where possible. expressed in seconds or in kilobytes transferred. 3DES in certain IOS versions.0(3)T). End with CNTL/Z. MD5) • authentication method – for identifying the routers attempting to establish a tunnel (choices: Rivest-Shamir-Adelman (RSA) signature.Advanced Security Services below with a short description of its purpose (the default setting is given first in all lists of choices): • priority number – a positive integer used to uniquely identify the policy when two or more are contained within the routers configuration file (default: none) • encryption algorithm – for protecting the IKE protocol messages (choices: DES. RSA encryption.536. • hash algorithm – for providing integrity to IKE protocol messages (choices: SHA. 12.

This parameter set can be built using the following three steps: 1. Creating the appropriate access lists Some administrators will want to create tunnels to protect all protocol data flowing between two routers. with ! the priority given to lower numbers Remote(crypto-isakmp)# encryption 3des ! If the user’s version of the IOS only supports DES.g. and http traffic). and ! community of interest data separation is needed.1c . all telnet. Remote(config)# crypto isakmp policy 1 ! The policy number may be an integer between 1 and 65. The following example displays an access list needed to protect ALL protocol information between the North and 208 Version 1. then use the ! following command to select DES for encryption ! Remote(crypto-isakmp)# encryption des Remote(crypto-isakmp)# hash sha Remote(crypto-isakmp)# authentication pre-share Remote(crypto-isakmp)# group 2 Remote(crypto-isakmp)# lifetime 86400 Remote(crypto-isakmp)# exit Remote(config)# exit Remote# Using the show crypto isakmp policy command in privileged EXEC mode (on the console of Remote or North) should now display the following information: North# show crypto isakmp policy Protection suite of priority 1 encryption algorithm: 3DES – Triple Data Encryption Standard (168 bit keys) hash algorithm: Secure Hash Standard authentication method: Pre-Shared Key Diffie-Hellman group: #2 (1024 bit) lifetime: 86400 seconds. We now need to give the tunnel its desired characteristics.536. ftp. one per line. no volume limit Default protection suite encryption algorithm: DES .Data Encryption Standard (56 bit keys) hash algorithm: Secure Hash Standard authentication method: Rivest-Shamir-Adleman Signature Diffie-Hellman group: #1 (768 bit) lifetime: 86400 seconds. End with CNTL/Z.Router Security Configuration Guide Remote# Remote# config t Enter configuration commands. Others will desire to protect only particular services or a subset of the data flow (e. no volume limit North# Establishing the IPSec Protection Parameters Using the pre-shared key and the security policy. IKE will determine preliminary information needed to create IPSec tunnels.

0 End with CNTL/Z.0.0. see Section 4.255. somewhat simplified.255 14. Access lists can be used to improve the granularity of the IPSec tunnels.0 0.255 14.0 Remote(config)# access-list 161 permit ip 14.1. Using the any option (e.0.0.0 0. End with CNTL/Z.0 0.255. North(config)# access-list 161 permit ip 14.255.0 209 .3 to learn more about access lists.2.0 0.255.255 Remote(config)# 2. If the tunnel needs data authentication protection. The syntax for an access list rule.2. access-list 161 below) for both the source and destination in the access list will force all packets to be IPSec protected.255 North(config)# and Remote# config t Enter configuration commands.2.255.0 North(config)# access-list 161 permit ip Choosing the any option for the source and destination also eliminates the need for netmasking in the access list. Configure the appropriate transform set The Cisco transform set identifies the desired protection mechanisms for building the IPSec tunnel.255.0 0.0.0. then choose the ESP protocol with either the DES or 3DES encryption algorithms (we highly suggest 3DES).g. one per line.255. then choosing either the Authenticated Header (AH) or the Encapsulated Security Payload (ESP) IPSec protocols with either hashing algorithms SHA or MD5 will suffice.0.255.0 0.0. If the tunnel you are setting up needs data confidentiality protection.255 Remote(config)# access-list 161 permit ip 14. access-list access-list-number {deny | permit} protocol source source-wildcard source-options destination destination-wildcard destination-options The network administrator for the North and Remote routers should enter the IPSec access list into their routers using the following commands in privileged EXEC mode: North# config t Enter configuration commands.255 North(config)# access-list 161 permit ip 14.0 A network administrator could argue that data authentication is not really needed for a protective tunnel between gateway routers since this property is normally obtained by Version 1.0.255 14.255 7.255 14. one per line.0.255 Remote(config)# access-list 161 permit ip 7.0.0 0.0.0. Remote(config)# access-list 161 permit ip 0.0.Advanced Security Services Remote routers.255 7.0.0 North(config)# access-list 161 permit ip 7. is shown below.

but leaves the original IP header intact. the transform-set used to build the tunnel (covered above in section 2).Router Security Configuration Guide an application behind the router which is pushing data through the tunnel. but adding it can improve defense in depth. The IPSec standards requires that tunnel mode be used when routers are employed as gateway security devices. the security association lifetime for the tunnel (in kilobytes and/or seconds). The transform set also specifies what part of each packet is protected by the IPSec tunnel. This mode protects both the original data portion of the IP packet and the original packet header. For a hostile network scenario. to which you can give a variety of transform-set related commands.1c . This mode protects the original data portion of the IP packet. transformN. IOS will enter crypto transform set configuration mode. the ESP protocol is chosen with both data protection and authentication properties applied to all information transmitted between the North and Remote routers. Configure the IPSec transform sets using the following commands: North# config t Enter configuration commands. one per line. one per line. In the following example. For more information on both the encryption and authentication algorithms. This information includes: the access-list to specify what traffic should be protected (covered above in section 1). consult the Cisco IOS 12. then the transport mode will be sufficient. Remote(config)#crypto ipsec transform-set set1 esp-3des esp-sha-hmac ! The name set1 is an arbitrary name Remote(cfg-crypto-trans)# mode tunnel Remote(cfg-crypto-trans)# end Remote# 3. The command syntax for defining an IPSec transform set is: crypto ipsec transform-set transform-set-name transform1 transform2 . When you give this command. . . Use exit to leave transform set configuration mode.0 Security Configuration Guide [2]. End with CNTL/Z. and 210 Version 1. North(config)# crypto ipsec transform-set set1 esp-3des esp-sha-hmac ! The name set1 is an arbitrary name North(cfg-crypto-trans)# mode tunnel North(cfg-crypto-trans)# end North# and Remote# config t Enter configuration commands. and creates a new IP header using the routers’ IP addresses. the remote address for the peer end of the IPSec tunnel. End with CNTL/Z. Create the necessary crypto map Cisco IOS uses crypto maps to bring together all information needed to create IPSec tunnels. If the IPSec tunnel is used for separating communities of interest over a protected network. and the tunnel modes. the preferred mode is tunnel (which is the default). This hides potentially sensitive IP protocol information about the networks and applications that are sending data through the tunnel.

one per line. they limit the length of time and ! number of bytes the tunnel is good for data protection before ! automatic rekeying occurs Remote(config-crypto-map)# set security-assoc lifetime kilo 80000 Remote(config-crypto-map)# set security-assoc lifetime sec 26400 Remote(config-crypto-map)# exit Remote(config)# exit Remote# Version 1. one per line. those with lower the sequence numbers have higher priority. See “Configuring IPSec Network Security” in the Cisco IOS 12. Each crypto map is identified by a map-name and a positive integer sequence number (called seq-num below).1c 211 . they limit the length of time and ! number of bytes the tunnel is good for data protection before ! automatic rekeying occurs North(config-crypto-map)# set security-assoc lifetime kilo 80000 North(config-crypto-map)# set security-assoc lifetime sec 26400 North(config-crypto-map)# exit North(config)# exit North# and Remote# Remote# config t Enter configuration commands.12. The syntax for the crypto map command is: crypto map map-name seq-num ipsec-isakmp.Advanced Security Services whether to use the IKE protocol in setting up the tunnel. Configure the IPSec crypto maps using the following commands: North# North# config t Enter configuration commands. which combine the previously entered configuration information.20 North(config-crypto-map)# set transform-set set1 ! The following are optional.1. North(config)# crypto map pipe-1 1 ipsec-isakmp ! The name pipe-1 is an arbitrary name North(config-crypto-map)# match address 161 North(config-crypto-map)# set peer 7. If two or more crypto maps with the same name are used. while the sequence numbers are used to set the priority for two or more crypto maps with the same name. End with CNTL/Z. The map-name used can represent one or more crypto maps. Remote(config)# crypto map pipe-1 1 ipsec-isakmp ! The name pipe-1 is an arbitrary name Remote(config-crypto-map)# match address 161 Remote(config-crypto-map)# set peer 14. The following example shows the construction of a single crypto map for the North and Remote routers.0 Security Configuration Guide to learn more about crypto maps. End with CNTL/Z.20 Remote(config-crypto-map)# set transform-set set1 ! The following are optional.0.2.

If both routers are connected to the Internet.1c . As a quick check (which could potentially eliminate many headaches) before turning on IPSec. End with CNTL/Z.20 on North should. North# config t Enter configuration commands. Assuming the ping was successful. in all likelihood. give us this answer.1. then remove the crypto maps from the interfaces which they were applied. one per line. one per line. End with CNTL/Z. we are now ready to build a tunnel between our routers. without an IPSec tunnel). North(config)# interface ethernet 0/0 North(config-if)# crypto map pipe-1 North(config-if)# end North# and Remote# config t Enter configuration commands. End with CNTL/Z. A simple ping 7.12. then the following commands should prepare both routers to establish an IPSec tunnel at the first beckoning of an IP packet which matches access lists 161. using outside interface eth0/0. make sure the two routers are in a state where they can communicate (i. North# config t Enter configuration commands.12.Router Security Configuration Guide The command show crypto map will display the following information on the North router (assuming no other crypto maps have been entered): North# show crypto map Crypto Map "pipe-1" 1 ipsec-isakmp match address 161 peer 7. Remote(config)# interface ethernet 0/0 Remote(config-if)# crypto map pipe-1 Remote(config-if)# end Remote# If IPSec is no longer needed to protect traffic between two routers.e.20 set transform-set set1 set security-association lifetime kilobytes 80000 set security-association lifetime seconds 26400 North# Turning on IPSec at the Appropriate Interface Once the previous steps have been completed. North(config)# interface ethernet 0/0 North(config-if)# no crypto map pipe-1 North(config-if)# end North# and 212 Version 1. one per line. as shown below. as in Figure 4-1. we are almost ready to build a tunnel between the North and Remote routers.1.

20 14. #pkts encrypt: 5. End with CNTL/Z.0.2. the first ping requests might time out since the computation time needed for the IKE key exchange / IPSec computations varies depending on the size of the router.20/255.e. #recv errors 0 local crypto endpt. etc. Once the IPSec tunnel has been established. If everything has been set up properly. local addr.1. These parameters can be seen using the show crypto ipsec security-association and the show crypto isakmp securityassociation commands.12.255. 14.255.: 7.} #pkts encaps: 5.1.255.: 14.Advanced Security Services Remote# config t Enter configuration commands. the transform set).1.2. Depending on the time allotted for a ping echo reply to return to the ping source.2.2.255/0/0) remote ident (addr/mask/prot/port): (7. If the negotiation is successful.0. one per line.20/255. the user should be able to review the IPSec tunnel parameters.12.20 PERMIT. the access lists will have notified the IOS that an IPSec tunnel has been requested to protect packet data.255/0/0) current_peer: 17.12. #pkts decrypt: 5.255. #pkts verify 5 #send errors 5.20 path mtu 1500. Remote(config)# interface ethernet 0/0 Remote(config-if)# no crypto map pipe-1 Remote(config-if)# end Remote# Testing A quick way to test if our IPSec tunnel has been established between the two routers is to simply execute a ping from one router to the other.20 QM_IDLE conn-id 1 slot 0 North# show crypto ipsec sa interface: Ethernet0 Crypto map tag: pipe-1. flags={origin_is_acl.0.20 remote crypto endpt.1c 213 .12. North# show crypto isakmp sa dst src state 7.20 local ident (addr/mask/prot/port): (14.0. #pkts digest 5 #pkts decaps: 5. This will cause the routers to use the IKE protocol (including the IKE authentication key and the IKE security policy information) for authenticating the two routers and facilitate the negotiation of the IPSec tunnel’s protection algorithms (i.1. speed of the network. the tunnel will be established and the ping requests will be protected. media mtu 1500 current outbound spi: 1B908AE inbound esp sas: spi: 0xEFA038E(251265934) Version 1.

Re-initialize the IPSec parameters by removing the IPSec and IKE security associations When an attempt is made to construct an IPSec tunnel between two peers. 214 Version 1. 1. Hence. If any one of many parameters are not set properly. The EXEC mode commands clear crypto sa or clear crypto isa commands. the following suggestions will help reset the IPSec relevant router parameters and hopefully allow for a tunnel to be constructed. interface eth0/0 on both North and Remote above). crypto map: pipe-1 sa timing: remaining key lifetime (k/sec): (4607999/3459) IV size: 8 bytes replay detection support: Y inbound ah sas: outbound esp sas: spi: 0x1B908AE(28903598) transform: esp-3des esp-sha-hmac . several commands may be used to collectively remove the unwanted information.g. } slot: 0. } slot: 0. crypto map: pipe-1 sa timing: remaining key lifetime (k/sec): (4607999/3459) IV size: 8 bytes replay detection support: Y outbound ah sas: Troubleshooting Most current IPSec implementations. this information will reside in IOS memory and hinder future attempts at constructing tunnels between these two peers. To remove this information and allow the routers to begin a fresh IPSec negotiation of tunnel parameters. including Cisco’s. in use settings ={Tunnel. then the information will be removed. all tailored to the specific peer devices involved. First. if the crypto maps are removed from the interfaces where they were placed (e. a few Cisco IOS releases have demonstrated unstable functionality: in some cases packets which should be protected by the tunnel are passed in the clear. can be very temperamental. conn id: 3. and the global configuration mode command no crypto ipsec sa. If the tunnel fails to be constructed.Router Security Configuration Guide transform: esp-3des esp-sha-hmac .1c . several things can be done. If the crypto maps are in use by established tunnels. in use settings ={Tunnel. then removing them is not a viable option. will remove the unwanted information. the IOS stores certain information about both of their IPSec configuration files. And even when a tunnel is established. the construction of the IPSec tunnel will not succeed. conn id: 2. If your routers do not correctly establish the IPSec tunnels that you need.

but both were on while the IPSec tunnel was being constructed. then the following access lists on Remote and North would satisfy the mirror access list requirement and should allow for the tunnel to be constructed between North and Remote. On the other hand.20 0.0. Make sure the routers have mirror access lists The Cisco IOS IPSec code can get easily confused when the access lists. Turning on the debug commands to observe the router’s IPSec negotiation It can be very helpful to run both the debug crypto ipsec and the debug crypto isakmp commands.12.0.29 (I) QM_IDLE 00:19:35: ISAKMP (1): received packet from 7. if we only want to protect packets to/from a LAN behind the Remote router (IP address 7.Advanced Security Services 2. M-ID of 405257172 00:19:35: ISAKMP (1): sending packet to 7. (Note: These debug options were run at different times. On North: access-list 101 permit ip 14.1/24) with anyone behind the East router (IP address 14. and determine exactly where the negotiations are failing. (Note: If the routers establishing the IPSec tunnel are not currently operational. Below is a list of the North router’s output when these two debug commands were turned on.20 (I) QM_IDLE Version 1.255 On Remote: access-list 102 permit ip 7.12. round-trip min/avg/max = 32/33/36 ms North# 00:19:35: ISAKMP (1): beginning Quick Mode exchange.) North# debug crypto isakmp Crypto ISAKMP debugging is on North# ping 7.!!!! Success rate is 80 percent (4/5).255 14. In our example above.1.0.255 3.) The debugging messages will allow the network administrator to observe how the local router is processing the remote router’s IPSec packets during the tunnel negotiation.20 Type escape sequence to abort.1.20.0. timeout is 2 seconds: .20 0. Full debugging imposes too great a load to be practical for operational routers. which are engaged by the crypto maps to determine what packets are protected using the IPSec tunnel.20/16). which can be entered while the router is in privileged EXEC mode.1. are not mirror images of each other.1. we can see that the access lists used by both North and Remote are mirror images since they both involve using the any option to indicate that all protocol packets. with source and destination addresses each behind one of the routers.1 0. turning on full debugging using the debug all command supplies even more diagnostic information.1c 215 . get protected.1. 100-byte ICMP Echos to 0. Sending 5.255

2. transform= 3esp-des esp-sha-hmac .255/0/0 (type=1).20 to 7.12.20/255. conn_id= src_proxy= 14.255.255/0/0 (type=1). (key eng.255.20.. keysize= 0.20 for prot 3 4w0d: IPSEC(key_engine): got a queue event. src= (key eng.12. ESP_3DES ISAKMP: attributes in transform: ISAKMP: encaps is 1 ISAKMP: SA life type in seconds ISAKMP: SA life duration (basic) of 3600 ISAKMP: SA life type in kilobytes ISAKMP: SA life duration (VPI) of 0x0 0x46 0x50 ISAKMP: authenticator is HMAC-SHA ISAKMP (1): atts are acceptable. spi= 0x0(0).12. src_proxy= 14.0.1. protocol= ESP.2. dest_proxy= 7.20 to to 14.0. lifedur= 0s and 0kb.12.20/255.Router Security Configuration Guide 00:19:35: 405257172 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 0x0 00:19:35: 00:19:35: 00:19:35: 405257172 00:19:35: 405257172 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: 00:19:35: ISAKMP (1): processing SA payload.20.2.0. msg. message ID = ISAKMP (1): processing ID payload.1. protocol= ESP.255.) dest= 7. 4w0d: IPSEC(initialize_sas): .0..1c .255.255.20 ) has spi 595658916 and conn_id 5 and flags 4 lifetime of 3600 seconds lifetime of 4608000 kilobytes ISAKMP (1): sending packet to 7.12.1. 4w0d: IPSEC(spi_response): getting spi 595658916 for SA from 14.20 ) has spi 59056543 and conn_id 4 and flags 4 lifetime of 3600 seconds lifetime of 4608000 kilobytes outbound SA from 14. dest_proxy= 7.20 to 7. lifedur= 3600s and 4608000kb.20 (proxy (type=1).255/0/0 (type=1).2.0.. message ID = ISAKMP (1): Creating IPSec SAs inbound SA from 7..1.1.20 (proxy 7. flags= 0x4 4w0d: IPSEC(key_engine): got a queue event.20 to 7. ISAKMP (1): processing NONCE payload.) dest= 7. src= 14.1. msg. (I) QM_IDLE North# no debug all North# debug crypto ipsec Crypto IPSEC debugging is on North# ping 7.1.255. 216 Version North# 4w0d: IPSEC(validate_proposal_request): proposal part #1.20/ transform= 3esp-des esp-sha-hmac .20/255.12. message ID = ISAKMP (1): Checking IPSec proposal 1 ISAKMP: transform 1.20.1.

255/0/0 (type=1).2. and for recovering from failures. sa_trans= 3esp-des esp-sha-hmac . spi= 0x385219F(59056543).2. Another use of IPSec would be to use it to protect the administration of a Cisco router. sa_conn_id= 100 4w0d: IPSEC(create_sa): sa created. flags=0x4 4w0d: IPSEC(create_sa): sa created. 5. then examine the configuration sequence for a PC running Microsoft Windows 2000 or XP. conn_id=101. src= 14.Advanced Security Services spi= 0x238108A4(595658916). or isolating networks within an organization. sa_spi= 0x385219F(59056543).20. Let’s say the computer the administrator uses to configure the router has IP address 14.6. keysize=0.2.0. msg. IPSec can be used to encrypt this communication. (sa) sa_dest= 7. perform the following steps: 1.255. transform= 3esp-des esp-sha-hmac .255.255.12. so we’ll secure a connection to there.20/255.2.2. The IP address of the North router on the interface closest to the administrator is 14.1. sa_conn_id= 101 North# no debug all 4. sa_trans= 3esp-des esp-sha-hmac . sa_prot= 50. protocol= ESP. conn_id=100.12.250.flags=0x4 4w0d: IPSEC(initialize_sas): .0.20.20. First. sa_prot= 50. (sa) sa_dest= 7. a computer on the desk of the administrator is to be used to administer the North router. dest_proxy= 7.255. This could represent either connecting widely separated networks.255/0/0 (type=1). Using IPSec for Secure Remote Administration The example used throughout the preceding section was to securely connect two networks from their gateways (which were Cisco routers).12.2.1c 217 . Common ways to perform administration of a Cisco router are to use either telnet (which sends the password in the clear) or SNMP.1.12. lifedur= 3600s and 4608000kb.20/255.) dest= 7. Enter configuration mode: Version 1.1. we’ll set up the configuration on the router.1. eliminating the threat of a network sniffer seeing passwords or sensitive configuration information. src_proxy= 14.9. is invaluable in diagnosing exactly where the tunnel negotiation is failing. like that obtained from running the debug commands on the router. (key eng.1. keysize=0. Since both of these run over IP.20. Use an IP packet sniffer to observe the contents of each packet in the IPSec tunnel negotiation This information. Configuring a Cisco Router for IPSec Secure Remote Administration On the Cisco router. which is next to the servers in Figure 41. In this example. sa_spi= 0x238108A4(595658916).

if some form of a public key infrastructure (PKI) is already in place. The key exchange size of group 2 is larger than that for group 1. and punctuation (see Section 4. DES has been demonstrated to be weak. North(config)# crypto isakmp key my4pa$$phra$eHere address 14. The default hashing algorithm. we can use transport mode instead of tunnel mode. The encryption options are DES and 3DES. Enter the authentication password. End with CNTL/Z. certificates just for this probably aren’t warranted.1c .9. However. so again we select the stronger option. The policy number selected here is 10.Router Security Configuration Guide North# config t Enter configuration commands. and so pre-shared keys can be used. and possibly not even strong enough to protect passwords. and SHA.2. Pre-shared keys are passwords – or better yet a passphrase. so we recommend 3DES.e. numbers. certificates can be used. which is just an arbitrary number to set a priority. North(config)# no access-list 12 North(config)# access-list 12 permit 14. North(config)# crypto ipsec transform-set 3des-sha-xport esp-3des esp-sha-hmac North(cfg-crypto-trans)# mode transport North(cfg-crypto-trans)# exit North(config)# 218 Version 1. The other option is the lifetime until a key renegotiation is required. this does not concern us too much. so we will take the default and not enter it. we want to use 3DES. one per line. SHA. North(config)# crypto North(config-isakmp)# North(config-isakmp)# North(config-isakmp)# North(config-isakmp)# North(config)# isakmp policy 10 authentication pre-share encryption 3des group 2 exit 4. Please do not use anything in the dictionary. include letters. it is not forwarding this particular traffic anywhere else).6 North(config)# line vty 0 4 North(config-line) access-class 12 in North(config-line) exit North(config)# 3. Create an ISAKMP policy. Since the authentication is only between 2 machines. Enable telnet access to the router for administration from the administrator’s machine.2. or anything easily guessed. The transform-set contains the parameters for protecting the actual traffic. if two or more ISAKMP policies exist on North. North(config)# 2.1. so we will skip this. Since we are treating the router as just a host to connect to (i. Again. This choice is also made because currently it is easier to configure IPSec in Windows 2000 to use transport mode.9. We’ll use access list 12 to list the machines that may to telnet to the router. is suitable. but again.5 for more guidelines on password quality).6 5.

The IPSec connections must be allowed.6 host 14. “Add” the IPSec snap-in for the local machine. Click the “Add” button and you will see all the possible snap-ins.6 North(config)# access-list 167 permit ip host 14.1. Finally. 167. we’ll ensure that no other crypto maps are still in existence before we define this one. North(config)# interface ethernet 0/1 North(config-if)# no crypto map North(config-if)# crypto map cisco-admin North(config-if)# exit North(config)# exit North# Configuring Windows 2000/XP for Secure Remote Administration Once the Cisco router has been set up.250 interface is named Ethernet 0/0). and “OK” the one snap-in that you have added. That will give you a window containing the currently added list of snap-ins. North(config)# crypto map North(config-crypto-map)# North(config-crypto-map)# North(config-crypto-map)# North(config-crypto-map)# North(config)# cisco-admin 10 ipsec-isakmp set peer 14. the Windows 2000 computer on the desktop of the administrator can be prepared.1. or by using the “Run” command from the “Start” menu). When doing so. This section assume moderate familiarity with Windows 2000 or XP network administration. Then we exit from configuration mode. run Microsoft Management Console (MMC).9. North(config)# access-list 167 permit ip host 14. We number the access list as 167. and IPSec should be running on the Cisco router. The match address links the desired access-lists to the crypto map. Version 1.1. “Close” the list of additional snap-ins. It will ask which computer it should manage. click “Finish”.250 host 14. The management console window should now look something like the screenshot below.1c 219 .) First. apply these definitions to the interface (the 14. Scroll down until you see one titled “IPSec Policy Management” or “IP Security Policy Management” and select that one. so we use the one we entered in the previous step. A crypto map must be created. You can add the snap-in by looking under the “Console” menu and selecting “Add/Remove Snap-in”. but the directions also apply to Windows XP.2.9.2. and select “Local Computer”.2.1. Any name can be given to this – we use ciscoadmin.2. either from a command window prompt. Priority for this crypto map is set to 10.250 7.1. A few of the dialogs may look slightly different. initially empty.Advanced Security Services 6. (The screenshots shown below were captured on Windows 2000.6 set transform-set 3des-sha-xport match address 167 exit 8.9.

Click “Next”. for which we will use the Add Wizard. and you aren’t required to fill in a description. and then hit the “Finish” button. In this window. but before that. ensure that the “Edit properties” box is selected. which we will do in a second. perhaps something like “Admin to Router”. Click “Next”. Any name will do. A wizard shows up to assist you on this quest. A new rule must be added. click “Next”. Click so that the default response rule is not activated. It asks for a name and description for this new policy.Router Security Configuration Guide Click right on “IP Security Policies on Local Machine” (either the left or right window will work) and select “Create IP Security Policy”. we will configure the key exchange parameters (which were called by the name isakmp in the 220 Version 1. The following window should appear. Two things must be done in this window.1c .

either modify a currently existing method by highlighting one and hitting the “Edit” button. Now you will see the following new window. If such a method does not exist. In this window. In the tabs at the top of this window. under “Protect identities with these security methods”. and any values for when to rekey are acceptable. Those values are IKE negotiation (Cisco calls it ISAKMP. Use the sideways scroll bar to see if a security method exists with the same settings as on the router. which is the 1024 bit Diffie-Hellman option. SHA1 Integrity (the hashing algorithm).Advanced Security Services Cisco configuration). or click the “Add” button to create a new one. To ensure everything is set up the same as on the Cisco. do not check the “Master key Perfect Forward Secrecy” button. Click that. The window that appears contains the title “Key Exchange Settings”. and “Medium (2)” for the Diffie-Hellman size (which is Group 2. In either case. Version 1. 3DES encryption. you probably should click on the correct one (which will highlight it). select “General”. Under that tab at the bottom of the screen is a button for “Key Exchange using these settings” with the word “Advanced” written on the button. The others can either be deleted or just left there. and click the “Move up” button until it is the first option on the list. not the 768 bit group 1).1c 221 . which is actually the name of a foundational specification on which IKE is built). click the “Methods” button.

1c . which will make the following tunnel endpoint window appear. Continue on without specifying a tunnel. When the introduction screen for the wizard shows up. Click the “Add. we do not need this tunnel.. Click “Next”. Recall that we previously selected “my4pa$$phra$eHere” as our choice when we configured the Cisco router.Router Security Configuration Guide Click “OK”. The next screen is about which network connections to use. which will use a second wizard. You should now be back at the window where you selected the “General” tab. click “Next”. Since we selected transport mode when configuring the Cisco router. or even better “All network connections” can be used.. Now is the time to enter the passphrase. choose either LAN connection. and then click it again on the next window. 222 Version 1.” button. but in this case. Now select the “Rules” tab and let’s continue. The network type “Remote access” is useful if you are using phone lines to connect remotely.

and click “Next”. The IP Filter List window will appear. From there. If you use the wizard. unselect the “Use Add Wizard”. but before you “Add”. So. it is probably empty. Version 1.1c 223 . click “Add” and the following IP Filter List definition window will appear. or something like that). This third wizard is not helpful. unselect the “Use Add Wizard” option. Initially. you get several screens in which you will type in the information you can supply to the one screen you see if you do not use the wizard. Name this filter (Cisco Only Filter. Now we need to add a filter.Advanced Security Services We enter that in the appropriate box. click “Add” and you should see the following screen.

Click OK. you will want to examine it in a bit more detail to be sure that it contains the options you need. Before selecting the “Require Security” option.255. Double click on “Require Security” to see what options are set.Router Security Configuration Guide You want it to have the Source address as “My IP Address” and the Destination address as “A specific IP Address” in which you fill in the IP address of the Cisco router. You do need to leave the mirrored option on so filters are defined for traffic going in both directions.2. “Request Security” and “Require Security”. 14. which you should “Close. Use a subnet address of 255. returning you to the filter list window.250.1c . “Permit”.” Then select that filter (call it “Cisco Router Filter”) from the list of filters and click “Next”.255. 224 Version 1. There are three default filters defined.1.255 which permits secure connections only to the one router and leaves all other communications unaffected. The next window that appears is the Filter Action window. It should look something like this.

The value in the third column should change to “Yes” and the policy will be imposed. We do not want to select the final two options. and select “assign” from the popup menu. This is not the case here. typically you must explicitly assign a policy after creating it. if you want to delete all but the one offer that is used. so we require it. “Assigned”. and that new policy will appear in the management console window. click “OK”. of the policy listing in the management console window. communication to other places is not affected and so not IPSec protected. “Allow unsecured communications with non IPSec aware computer” and “Session key Perfect Forward Secrecy”. and are not using the AH protocol. We want to select “Negotiate security” here. IPSec will be activated automatically. which is mostly used when the initial key exchange is shared. but always respond using IPSec”. that would not be bad. Look at the third column. you have created an IP Security Policy. In fact." The next time you connect to the Cisco router. The reason we don't want to allow unsecured communications is that this IPSec configuration only applies to communication with the router. When all these settings are correct.Advanced Security Services Click on the security method preference order options and edit them to ensure that at least one of them contains the cryptographic settings for protecting the actual data that was configured in the Cisco. we want to use security. For just this connection. Version 1. After following all these steps. Make sure that the policy is actually in effect. so any settings for that are acceptable. The only remaining thing to do is to click "Finish. Choose “Accept unsecured communication. The lifetime (until keys are renegotiated) is not important. and the traffic will be IPSec protected. we are using ESP with both 3DES and SHA.1c 225 . Highlight the “Require Security” button. and click “Next”. For our example. If the column contains the word “No”. then right-click on it. Perfect Forward Secrecy is a way to do a second key exchange.

226 Version 1. On the router. use the command show crypto ipsec sa to confirm that IPSec is being used.Router Security Configuration Guide A quick check to ensure that it is working is to ping the router from the Windows 2000 host. A network sniffer can be used to verify that communications between the router and host are encrypted.1c . and the router and the Windows 2000 host should have completed their key exchange and the ping should succeed. The first attempt should fail and report "Negotiating IP Security". Ping a second time.

9. The detailed example below shows how to configure the router North (14. Using SSH for Remote Administration Security An alternative to setting up IPSec for secure remote administration is to configure your router to use the secure shell service.2. Configuring a Router for Secure Remote Administration with SSH While in IOS enable mode. See Section 4. Configure an access list permitting access from the administrative host. Only certain Cisco IOS versions are shipped with the SSH feature set. With SSH. For a thorough discussion of SSH. Enter configuration mode: North# config t Enter configuration commands. and rcp services.2. For more information about access lists. Before you can configure SSH. see Section 8. North(config)# no access-list 12 North(config)# access-list 12 permit host 14. there are two important prerequisites to address.4 can support SSH version 2. Because the connection is encrypted. see Section 4.6 for detailed information on how to define usernames. perform the following steps: 1.1.1 log North(config)# line vty 0 4 North(config-line)# access-class 12 in North(config-line)# exit Version 1. SSH versions 1 and Cisco IOS 12. one per line. There are two main versions of the SSH protocol in widespread use. make sure that the router has a local hostname and domain name set. For more information about IOS versions. North(config)# 2.250) to run the SSH server and accept incoming connections.0 through 12. IOS 12.3.3. consult [13].1. End with CNTL/Z. commonly called SSH.Advanced Security Services 5.3.2. It utilizes RSA public key cryptography to establish a secure connection between a client and a server. 5. you must establish usernames for people that will be attempting to connect to the router. rsh. Versions after and including 12.1 to connect to the router for administrative purposes.1c 227 . rlogin.3 are currently capable of supporting only SSH version 1.3. In this scenario the administrator will use an SSH client on the host 14. passwords and other sensitive information are not exposed in the clear between the administrator’s host and the router. SSH also prevents session hijacking and many other kinds of network attacks. This example uses standard IP access list 12 to identify the hosts that may start SSH sessions into router North. IOS versions that do not support IPSec usually do not support SSH either. First.0(5)S with IPSec include support for SSH. Instructions for this can be found in section 4. SSH was originally intended to be a secure replacement for classic telnet.

.2.Router Security Configuration Guide 3. To change this from the default. If the router already has a key pair. Otherwise. How many bits in the modulus [512]: 2048 Generating RSA Keys . then you may use it for SSH. you may skip this step. because there is no way to recover a private key that has been deleted. If you need to remove an old key pair. then you may delete them using the command crypto key zeroize rsa. standard vty timeout settings apply. Choosing a key modulus greater than 512 may take a few minutes. but it will not be used until you configure it.2 for direction. The default authentication timeout is 120 seconds. North(config)# ip ssh time-out 90 North(config)# 228 Version 1. Below are some useful commands for further configuring the new SSH server. as specified in section 4. If you delete the router’s RSA key pair. generate a new RSA key pair for this router.) Cisco suggests a minimum modulus size of 1024 bits.6. Once the connection is established. If you have already created user accounts (with or without AAA). (Note: you must assign a domain name before creating a key. the SSH service will be present on the router whenever an RSA key pair Choose the size of the key modulus in the range of 360 to 2048 for your General Purpose Keys. The recommended value is 90. which is also the maximum allowed value.1c . do the following. see Section 4. and you are absolutely sure that the keys are not being used. This is the number of seconds the server will wait for a client to respond with a password. North(config)# crypto key generate rsa The name for the keys will be: North. perhaps generated as part of its IPSec configuration. To generate a new key pair. [OK] North(config)# If this command succeeds. the router must possess an RSA key pair. the SSH server is enabled and running. North(config)# username joeadmin password 0 1-g00d-pa$$word North(config)# line vty 0 4 North(config-line)# login local North(config-line)# exit North(config)# To act as an SSH server. Set up a username that is permitted to connect to the router. then the SSH server will stop. use the command crypto key generate as shown below. Note: check carefully before deleting a key pair.. • Configure an authentication timeout.dod. as detailed below. By default.

The default value is 3 attempts.3. Authentication retries: 2 North# To verify that SSH has been successfully enabled and check that your session is actually using SSH. the example below shows how to set the router to drop the connection at the second failure. simply leave off use the transport input command as shown below.1 To verify that SSH has been successfully enabled. connect to the router using your SSH client and type the command show ssh.1c 229 . these commands vary between IOS releases. North# show ip ssh Version 1.2.Advanced Security Services • The number of incorrect login attempts that are permitted before the router will drop a remote access connection is also configurable. Advanced SSH Commands There are a number of commands that will allow you to verify that the SSH server is now operational. execute the following command.5 North# Encryption State 3DES Session Started Username joeadmin IOS Version 12. but if an IOS router supports SSH then Telnet should not be used.2 To verify that SSH has been successfully enabled. If your session is secure then the output should resemble that shown below. and verify your output is similar to the following. which is a sound choice. and verify your output states that SSH is enabled. the maximum value is 5. There may or may not be any current connections. execute the following command. It is also possible to accept both SSH and Telnet. However. The two subsections below describe the commands in detail. North(config)# line vty 0 4 North(config-line)# transport input ssh North(config-line)# exit North(config)# 5. North(config)# ip ssh authentication-retries 2 North(config)# To disable telnet and require SSH. Do not set the value higher than three. which is recommended. depending on how you are connected to your router. North# show ssh Connection Version 0 1. IOS Version 12. North# show ip ssh SSH Enabled – version 1.5 Authentication timeout: 90 secs.

1.Router Security Configuration Guide Connection 0 North# Version 1.1c . North# ! enable debug messages from the SSH service North# debug ip ssh North# ! disable debug messages from the SSH service North# no debug ip ssh A Sample SSH Session The sample session below shows how to connect from a Unix host to a Cisco router using the OpenSSH client. Obtain the connection number using one of the two methods shown above.5 3DES Session started joeadmin North# exit Connection to 14.2. % ssh –l joeadmin 14. which is in use by user joeadmin. This example disconnects session 0.5 Authentication timeout: 90 secs.250 joeadmin@14.1.250 closed. North# disconnect ssh 0 North# . It is very important to disable debug messages when you are finished using them. % 230 Version 1. there are two ways to go about this. then use of the commands below to disconnect the session.250’s password: Warning: Remote host denied X11 forwarding North> enable Password: North# show ip ssh SSH Enabled – version 1.1. or that a connection must be forcefully closed.or North# clear line vty 0 North# You can use the IOS command debug ip ssh to diagnose SSH operation.5 Encryption 3DES State 4 Username joeadmin Diagnosing and Managing SSH In the unlikely event that a connection has not properly closed. Authentication retries: 2 North# show ssh Connection Version Encryption State Username 1 1.1.1.

Unix / Linux • OpenSSH (freeware) • SSH Secure Shell (commercial) Windows • PuTTY (freeware) • TTSSH Plugin for TeraTerm Pro (freeware) • SecureCRT (commercial) • SSH Techtia client (commercial) MacOS • NiftyTelnet 1.version 2. North(config)# ip ssh version ? <1-2> Protocol version North(config)# ip ssh version 2 North(config)# exit North# show ip ssh SSH Enabled .4.Advanced Security Services 5. Security of SSH Versions There are several known security weaknesses with the SSH version 1 protocol. When your IOS release supports SSH version 2.3. There are usable clients available for almost every host platform (even PDAs).1 (freeware) 5. Below are some useful clients for common host operating systems. and most Unix and Linux distributions include an SSH client by default.3.3. Even though SSH version 1 may be subject to network man-in-the-middle attacks in some circumstance. the host must support an SSH client. but tools that implement some of them do exist. Information on where to download these applications can be found in the Tools listing. Some Available SSH Clients To employ SSH between an administrative host and the router.1c 231 .3. For more information about vulnerabilities in SSH and Cisco IOS. Section 9. configure the router to use it exclusively as shown below. consult [12]. Authentication retries: 2 North# Version 1. it is still a much more secure choice for remote administration than unprotected Telnet. Attacks that exploit these weaknesses are complex and non-trivial to execute.0 Authentication timeout: 90 secs.

” Much of the documentation still uses the old name. Basic Concepts A network firewall is a network device that connects a protected internal network to some other untrusted. Configured together with good access lists. consult the Cisco IOS documentation for information about IOS NAT features. To reach even a moderate level of effectiveness as a firewall.Router Security Configuration Guide 5.) 5. This guide does not describe NAT. many router vendors.) 5.1c . Using NAT. while at the same time allowing authorized users on the trusted network (the ‘inside’) access to services on the untrusted network (the ‘outside’). but to offer no such services in the other direction. Cisco renamed the IOS Firewall to “Cisco Secure Integrated Software. The security policy for this setup is to permit users to take advantage of certain network services on the untrusted network. if it is running a version of IOS that has firewall capabilities. Note that CBAC is intended mainly for border routers.2. Current product catalogs and web pages use the new name. (Another important feature for firewalls is hiding network addresses and structure. The Cisco IOS Firewall feature set Content-Based Access Control (CBAC) facility allows a router to act as a rudimentary stateful inspection firewall. 232 Version 1. by transparently translating all IP addresses and coalescing distinct IP addresses into a single one. Stateful inspection firewalls do this by inspecting each packet for compliance with the specified security policy.4.4. As long as all traffic between the trusted and the untrusted network pass through the firewall. and to block attacks on your network.1. CBAC can provide modest firewall protection for a network without extra hardware. Configuring Cisco IOS Content Based Access Control The Cisco IOS Firewall feature set is designed to prevent unauthorized. it offers you another facility for enforcing security policy at the boundary between different networks.4. (Note: in mid-2000. possibly hostile network.3 describes access lists in detail. and a firewall between a main network and a restricted enclave. a router can hide the structure of the trusted network. and that is what we will use below. Using a Cisco Router as a Firewall This section describes how to use a Cisco router as a modest firewall. Cisco IOS provides full support for Network Address Translation (NAT). Potential applications for using a Cisco router as a firewall include: a modest Internet firewall. Because routers connect networks together. it can effectively enforce a number of network security capabilities. external individuals from gaining access to your internal network. a firewall between two different communities of interest. the router configuration must include good access lists. provide a rudimentary firewall capability in their routers. Section 4. including Cisco. The figure below shows the basic structure for a CBAC-based firewall setup.

CBAC inspects the initial TCP packet of the connection.10. permitting data from 7. 1.6 User Host 14.1c 233 . there must be at least one access list in place on the path from the untrusted network to the trusted network. The figure below illustrates this. or an outbound list on the inside interface. and use that information to make access decisions.64 Router 14. CBAC adjust access list 3. 3.0/24 User Host 14.10. inside interface inspect outside interface passes access list. 2.1. CBAC maintains connection state information for individual connections.7 Figure 5-1: A Simple Router Firewall CBAC examines not only network layer and transport layer information. Because CBAC works by modifying access lists.2. maintain state information.20 (port 80) on the untrusted network. either an inbound list on the outside interface.20 port 80. The decisions are enacted when CBAC dynamically adds rules to interface access lists to pass permitted traffic.2. and adds a rule to the inbound access list. but also examines the application layer protocol information (such as FTP information) to learn about the state of TCP and UDP connections. Outbound request Router Inbound response trusted network Host Figure 5-2: CBAC Overview Version 1.9. untrusted network 1.10. The heart of CBAC is the ability to inspect outgoing IP traffic in real-time.Advanced Security Services Untrusted Network 14. Host initiates a web connection to web server Trusted Network 14. Response comes back from the web server.2.2.

then CBAC is not supported. Central. then install an IOS version that does (see Section 4. OSPF. direct way to check whether a router has CBAC capability. follow these steps: 1. and a router with CBAC.1c . like H. Create a CBAC inspection ruleset supporting the desired services list. Call this list the desired services list. Set the CBAC global timeouts. Determine the list of services that users or hosts on the trusted network need from the untrusted network. 234 Version 1. permitting traffic that the router must process. Web (HTTP). must be separately permitted by the interface access lists if you need them. Example: FTP. but prohibiting other TCP and UDP traffic including the desired services list. Apply the CBAC inspection ruleset to an interface.Router Security Configuration Guide Note that CBAC handles only TCP and UDP protocols. It also includes some special case handling for multi-port application protocols. This is the access list that CBAC will be modifying on the fly. The easiest way to check is to execute a CBAC-related command. values that are too long can leave the trusted network vulnerable. usually the outside interface of a border router. South.5). There is no simple. Check that the router supports CBAC. 6. RealAudio (RTSP) 3. 7.0(9) with Firewall Feature Set 2. POP3. The two examples below show a router without CBAC. SMTP. and test it from the untrusted network by running a network scanner (see Section 6). if it does not. Set up an outbound access list on the outside interface. Steps in Setting Up a Cisco Router Firewall To set up a simple firewall using CBAC.5. Example: IOS 12. 4. or IPSec. such as ICMP. Other IP protocols and services.3). prohibiting all traffic that should not leave the trusted network but allowing traffic on the desired services list (see Section 4. Testing for CBAC Support on the Router Examine the router IOS installation to ensure it has the firewall feature set. Test the configuration from a host on the trusted network by running services. 8. 5. if the command fails. These timeout values determine the duration of window of accessibility opened back through the firewall in response to a request from the trusted network. Set up an inbound access list on the outside interface.323 and FTP. Step 1.

South# Step 2. NTP. and Telnet.1c 235 . plus SMTP and POP3 to the mail server only. Other UDP FTP Mail (SMTP) H. It also prevents use of non-standard ports for FTP data. FTP. Connect TCP protocol on port 25. etc. Control connection on TCP port 21. Generic UDP services. . H. Telnet. select ones to support by permitting them through the access list set up in Step 3. Because the Java blocking capability is very weak. CBAC in IOS 12. CBAC automatically tracks the RealAudio port assignments. IKE. CBAC has some ability to block Java applets. the most commonly used ones are listed below.323 (NetMeeting) RealAudio (RTSP) For web traffic (HTTP). Real-Time Streaming Protocol over UDP or TCP. Remarks CBAC will support any of these. For example. generic UDP must also be configured to use it. a reasonable list of desired services for many installations is: DNS. data on TCP port >1024. it is not typically employed. Central# versus South# show ip inspect all Session audit trail is disabled Session alert is enabled . POP3. TFTP. HTTP. NTP.323 video conference protocol over UDP. such as HTTP. Version 1. Determine the Application Services to Support Decide which application-layer protocols to permit using CBAC. CBAC permits only RFC 821 standard SMTP commands. select ones to support by permitting them through the access list set up in Step 3. CBAC will support any of these. etc.Advanced Security Services Central# show ip inspect all ^ % Invalid input detected at ‘^’ marker.0 supports about a dozen application service types. Because NetMeeting uses additional non-standard ports. Best practice on a router is deny all protocols except those identified as needed. and watches the FTP authentication exchange. SNMP. SSL. CBAC has special support for FTP. Service Basic TCP Protocols Definition Generic connected TCP protocols. such as DNS.

0.1c . or even both.255 any South(config-ext-nacl)# permit udp any eq ftp South(config-ext-nacl)# permit tcp 14.255 host 14.0.255 any eq ntp South(config-ext-nacl)# permit udp 14.0 0.2.2. even those on the desired services list.3). see Section 4. in general. the access list is applied to the outside interface.2.2. For more information about access lists. This access list must permit the protocols on the desired services list. this access list must be an extended IP access list.10.0 0.0. The example below shows an access list for our desired services list. this is a safe choice.2.0.10. Set up an Inbound Access List CBAC works by modifying inbound access lists: it can work with an access list applied to the interface on the trusted or untrusted networks.255 any eq telnet South(config-ext-nacl)# permit tcp 14.0.Router Security Configuration Guide Step 3.0 0.0.10. permits a modest set of useful ICMP messages. in the outbound direction.255 any eq www South(config-ext-nacl)# permit tcp 14.0. South(config)# ! create inbound access list for CBAC to work on South(config)# no access-list 111 South(config)# ip access-list extended 111 South(config-ext-nacl)# permit icmp any any echo-reply 236 Version host 14. the destination address can be the catch-all any.0.0. there must be an access list applied to traffic from the trusted net to the untrusted net.255 any eq domain South(config-ext-nacl)# permit tcp 14.0.0 0. The example access list below blocks TCP and UDP traffic effectively. Set up an Outbound Access List Before CBAC can do its work. South(config)# ! Create the access list South(config)# no access-list 110 South(config)# ip access-list extended 110 South(config-ext-nacl)# permit icmp 14.3 eq pop3 South(config-ext-nacl)# deny ip any any log South(config-ext-nacl)# exit South(config)# ! Apply the access list to the outside interface South(config)# interface eth 0/0 South(config-if)# ip access-group 110 out South(config-if)# exit South(config)# Step 4.2.3 eq smtp South(config-ext-nacl)# permit tcp 14.0 0.2.2. and permits the RIP routing protocol (see Section 4.9. In this example.0 0.0 Also.0. The source address for each rule in the access list should be a network address or address range valid for the trusted network. 0. An inbound access list intended for use with a simple CBAC firewall scheme should block all TCP and UDP services.0.

ip inspect name ruleset-name protocol [alert on/off] [audit-trail on/off] [timeout override-timeout] The alert option controls whether use of that protocol causes a console alert message to be generated. The name of the ruleset is “fw1. use the command ip inspect name. • The connection closes down (TCP only). similarly. beyond those generated by interface access lists. Enable the alert and audit-trail features to get additional log messages.1c 237 . South(config)# South(config)# South(config)# South(config)# South(config)# ip ip ip ip inspect inspect inspect inspect name name name name fw1 fw1 fw1 fw1 udp audit-trail on tcp audit-trail on ftp audit-trail on smtp audit-trail on Step 6. Adjust the CBAC Global Parameters When CBAC detects a connection attempt by a client on the trusted network. it adds a rule to the inbound access list to permit the expected response. and the second rule supports web. The syntax is shown below. • The connection is idle for longer than an allotted idle time. using the command ip inspect audit-trail. Telnet. (In older versions of CBAC.) The example ruleset below supports the example desired service list. This rule gets removed when one of the following conditions are satisfied: • The response does not arrive within the allotted timeout time. the audit-trail option controls whether use of that protocol causes a log message to be generated.” Its first rule supports DNS and NTP.Advanced Security Services South(config-ext-nacl)# permit icmp any any unreachable South(config-ext-nacl)# permit icmp any any ttl-exceeded South(config-ext-nacl)# permit icmp any any packet-too-big South(config-ext-nacl)# permit udp any any eq rip South(config-ext-nacl)# deny ip any any log South(config-ext-nacl)# exit South(config)# ! apply the access list to the outside interface South(config)# interface eth 0/0 South(config-if)# ip access-group 111 in South(config-if)# exit South(config)# Step 5. Create a CBAC Ruleset To create a CBAC ruleset. audit trails could only be turned on globally. and POP3 email services. Version 1.

Length of time CBAC continues to manage a TCP session with no activity.Router Security Configuration Guide The default timeout and idle times in Cisco IOS 12.0 are longer than necessary. The table below describes the parameters to change. Default 30 seconds 5 seconds Suggested 15 seconds 1 second TCP idle-time UDP idle-time 1 hour 30 seconds 30 minutes (1800 sec. The example below shows how to set the global timeout parameters. South(config)# interface eth0/0 South(config-if)# ip inspect fw1 out South(config-if)# end South# show ip inspect interface Interface Configuration Interface Ethernet0/0 . Length of time that CBAC continues to manage a UDP ‘session’ with no activity. synwait-time 15 finwait-time 1 idle-time 1800 idle-time 15 Step 7. End with CNTL/Z. . these values might need to be increased for a very slow connection (e. 238 Version 1.1c . but these can be left at their default values. South# config t Enter configuration commands. Use the interface configuration command ip inspect name to apply a ruleset. a modem) or on a highly congested network. Timeout Name Synwait-time Finwait-time Description Length of time CBAC waits for a new TCP session to reach established state. There are also global CBAC parameters related to half-open TCP session. one per line. Length of time that CBAC continues to manage a TCP session after it has been closed down by a FIN exchange. Apply the CBAC Ruleset to the Interface CBAC is not in force until a ruleset has been applied to at least one interface. The example below applies the ruleset from step 5 to the outside (untrusted network) interface.) 15 seconds Of course. South# End with CNTL/Z. South(config)# ip inspect tcp South(config)# ip inspect tcp South(config)# ip inspect tcp South(config)# ip inspect udp South(config)# exit South# one per line. South# config t Enter configuration commands.g.

2. to see that CBAC is working.9. see Section 6.4.Advanced Security Services After this step. and second. Test the CBAC Configuration Perform some simple tests from a host on the trusted network. Connected to 14.0.250).0.10. Telnet. SMTP (to a single host).0. Escape character is '^]'. Configuration Sample The configuration command listing below shows the configuration commands for a firewall router with a simple CBAC configuration.255 any permit udp 14. no access-list 110 ip access-list extended 110 permit icmp 14. The desired service list for this firewall is: DNS. 0.6) to a host on the untrusted network (14. Access is limited to authorized administrators only! Username: nziring Password: Central> While the Telnet session is active. starting a telnet session from a host on the trusted network to a host on the untrusted network. FTP.255 any eq ntp Version 1.. as illustrated in the example below. NTP.. and POP3 (to a single host).250. If the command gives no output.2. HTTP. This sample is formatted as it would appear in a configuration text file stored on a host for download to the router South. South# show ip inspect sessions Established Sessions Session 6187B230 (14.0.0 0. Step 8.1c 239 .10.2.250. The test shown here has two parts: first.2. For more detailed testing information. check the CBAC session status on the router using the command show ip inspect sessions.10.9. This is the CENTRAL router.250:23) tcp SIS_OPEN South# If the CBAC configuration seems to be working. confirming that CBAC is managing the session.2. save the router configuration to NVRAM at this point with the command copy running startup. CBAC should be running on the router. The example below shows a Telnet session from a host on the trusted network (14. then CBAC is not working.10.189:3175)=>(14.9. It should show the telnet session. 5.250 Trying 14. $ telnet 14. deny ip any any exit 0.9.1c . permit tcp 0.3 eq pop3 no access-list 111 ip access-list extended 111 deny ip permit tcp 14.0.255 0.0 permit tcp 14.255 any eq domain any eq www any eq ftp any eq telnet host any log ! permit routing updates permit udp any any eq rip ! permit useful ICMP message types permit icmp any any echo-reply permit icmp any any unreachable permit icmp any any ttl-exceeded permit icmp any any packet-too-big deny ip any any log exit ip ip ip ip ip ip ip ip inspect inspect inspect inspect inspect inspect inspect inspect name name name name tcp tcp tcp udp fw1 fw1 fw1 fw1 udp audit-trail on tcp audit-trail on ftp audit-trail on smtp audit-trail on synwait-time 15 finwait-time 1 idle-time 1800 idle-time 15 interface eth 0/0 ip access-group 110 out ip access-group 111 in ip inspect fw1 out end 240 Version 1.Router Security Configuration Guide permit udp 0.255 0.2.0 permit tcp permit tcp 14.0.255 0.0.255 0.3 eq smtp host 14.0.

only the first one found is reported by the IDS. and terminating unauthorized activity. if a packet is scanned and multiple. The 59 signatures are divided into two different categories: information signatures (“info”) and attack signatures (“attack”). or misuse in network traffic. 5. Attack signatures detect intrusions or attacks into the protected network. Both the firewall and IDS features should be enabled together for the best security. which could be a series of probes or access attempts distributed across multiple hosts over a random time period. it is important to note that a packet is subject to intrusion detection only if the router actually attempts to forward it. such as port scans and echo requests. The network administrator can then specify a particular action to take when an event signature is detected. The Intrusion Detection System on the router is a part of the IOS Firewall (CBAC) facility (Section 5. Cisco routers. The signatures were selected from a broad cross-section of intrusion detection signatures. Cisco IOS Intrusion Detection The Cisco IOS Firewall Intrusion Detection System (IDS) is a real-time IDS designed to enhance border router security by detecting. such as denial-ofservice attempts or execution of illegal commands during an FTP session. is that all network traffic flows through it and may be examined. Each packet is scanned against the Version 1. for example. Although the individual signatures cannot be modified. different signatures are detected. Packets dropped by an in-bound access list. Atomic signatures detect simple specific access attempts such as an attempt to access a specific port on a specific host. When using the IDS facility with other IOS security features. especially a border router. This facility is available in IOS releases for many. For adequate security. Information signatures detect many information-gathering types of activity. would not be scanned. reporting. which are either atomic or compound signature types. you can apply access lists to filter addresses and protocols from being subject to analysis by any particular signature. This feature is helpful in reducing false positives. Also. It is possible to enable them independently on separate interfaces. A unique benefit of implementing an IDS on a router.1. IOS IDS Basic Concepts Cisco has identified and incorporated into the IDS software 59 signatures of the most common actual and potential network attacks. but not all.1c 241 .4).5.5.Advanced Security Services 5. information-gathering scans. These signatures are used to match and detect patterns of security violations of the most common network attacks. Compound signatures detect more complex patterns. The Cisco IOS IDS is an in-line intrusion detection package that can monitor all packets and sessions flowing through the router. the IDS feature should not be configured as a standalone protection device. The two categories of information or attack signatures are also divided into two other categories.

3 for information on IOS versions. consult the “Traffic Filtering and Firewalls” section of the Cisco IOS 12. make sure that your router supports them by attempting to execute a simple IDS command.5. and the second the response from an IOS without IDS support (Central).Router Security Configuration Guide signature list in order to detect possible attacks or suspicious activity. Once you have determined that a particular router supports the IDS facilities. When an attack is detected. Central# If your router does not support the Firewall IDS facility. Configuring the IOS Intrusion Detection System Only those IOS releases marked “Firewall/IDS” support the IDS features described in this section.1c . one per line. the IDS logging facility. 5. 242 Version 1. The first example below shows the response to an IDS command from a version of IOS with IDS support (South). One way to initialize the facility is to set a parameter on one of the IDS signatures. drop the packet. Before attempting to configure the IDS features. South# versus Central# show ip audit all ^ % Invalid input detected at '^' marker.5 for information on loading IOS upgrades. depending on how the IDS was configured. After you’ve configured everything. This section presents only a brief overview of the IDS facility. The IOS documentation recommends using the command below. and Section 8. Step 1 .2 Security Configuration Guide (in the IOS documentation). South# config t Enter configuration commands.Initialization You must initialize the IDS facility before configuring it.2. South# show ip audit all Event notification through syslog is enabled Event notification through Net Director is disabled . follow the three steps outlined below to configure them. it is good practice to confirm that the IDS facility is working. First. See Section 4. for more details. initialize the IDS facility. . and/or reset a TCP session. it may be possible for you to upgrade your router to a release that does. Third. South(config)# ip audit smtp spam 200 End with CNTL/Z. the IDS will log an alarm to the syslog server or a Cisco Netranger Director. initialize the Post Office. configure and apply the audit rules. Second.

15 localaddress 14. Each alert must be logged if it is to be useful. and the org-id value should be the same as in command 2. The default queue size is 100. South(config)# ip audit notify nr-director South(config)# ip audit po local hostid 141 orgid 2 South(config)# ip audit po remote hostid 1 orgid 2 rmtaddress 14.2. set the size of the event queue for the Post Office. one per line.1c 243 . If the router has very little RAM (e. Next. ip audit po local hostid host-id orgid org-id This command sets the host and organization IDs for the Post Office. less than 32MB) then you should lower the value to 50. which is generally a reasonable setting. while the local-ip-address should be the IP address of the router interface closest to the Director.Advanced Security Services The value of 200 in this example is the maximum number of recipients that can be in an email message before the IDS will designate it as undesirable ‘spam’. You can also log the alerts to a syslog server (see Section 4. it throws an alert.10. The dir-ip-address should be the IP address of the IDS Director. the host-id must be a unique value for this router. and the org-id must be shared by the Director and all Cisco IDS sensors that send alerts to it.15. South# config t Enter configuration commands. above.2. South(config)# ip audit po max-events 50 South(config)# Step 2 .Configure the Post Office When the IDS facility detects a match to one of its intrusion signatures. If your network supports a Cisco Secure IDS Director. then configure the router to send the alerts to it.64 South(config)# Version 1. 3. The host-id value must be the host ID of the Director. This command can accept additional parameters. ip audit po remote hostid host-id orgid org-id rmtaddress dir-ip-address localaddress local-ip-address This rather complex command specifies the address and settings for sending alerts to the IDS Director. ip audit notify nr-director This command simply tells the router to use an IDS Director.g. Setting up the IDS facility to send alerts to an IDS Director requires three commands: 1.10. The example below shows how to configure the router South to send alerts to an IDS Director at 14.5 for information on syslog configuration). 2. also sometimes called a “Netranger director”. directing alerts is the job of the Post Office. see the IOS documentation for details. End with CNTL/Z.2.10.

although this will raise the performance burden imposed by IDS scanning. End with CNTL/Z. and may use standard IP access lists to limit the addresses to which the signatures are applied (see Section 4. The example below shows how to define a rule named IDR1 containing both info and attack signatures. . Configuring and Applying Audit Rules Once you have initialized the IDS and set up the Post Office. South(config)# ip audit name IDR1 info action alarm South(config)# ip audit name IDR1 attack action alarm drop reset South(config)# It is also possible to limit the application of a rule item with a standard IP access list. If you forget to do this. you can designate any or all of three possible actions to be performed when a signature is matched by traffic. or the actions taken. you must also configure the Director to accept alerts from the router.. you can restrict the detection of an event. South# config t Enter configuration commands. South# reload Proceed with reload? [confirm] y . one per line.discard the packet that triggered the alert • alarm . • drop . and then adding one or more specific items to the rule. Because 244 Version 1. the Director will not record the alerts sent by the router. An item must be based on either the “info” or “attack” signature classes.Router Security Configuration Guide Note that. Using an access list.5. South# copy running-config startup-config Building configuration.cut off the TCP session that matched the signature (TCP only) Use the config commands ip audit name to add items to an audit rule. you should configure the IDS facility to send alerts as normal IOS log messages using the command shown below.1c . you are ready to define audit rules and apply them to specific interfaces. South(config)# ! send IDS alerts to syslog and buffered log South(config)# ip audit notify log After setting or changing any Post Office parameters.3 for more information on access lists). If the network does not have a Cisco IDS Director available.log an alert to the Post Office and/or syslog • reset . you must save the running configuration and reboot the router. For each part of an audit rule. 5. after you have configured the router to send alerts to an IDS Director.3. You define an audit rules by choosing a name.

because all traffic received on that interface will be scanned. similar to the one described in Section 6.g.3. South(config)# South(config)# South(config)# South(config)# South(config)# ! don’t alert on ICMP source-quench ip audit signature 2002 disable ! don’t alert on ICMP time-exceeded ip audit signature 2005 disable You can apply a named audit rule on any interface. . consult the IOS documentation.1. Applying the rule in-bound will yield more complete scanning. Applying the rule outbound will reduce false positives.752 GMT: %IDS-4-ICMP_ECHO_SIG: Sig:2004:ICMP Echo Request .No connections are active South# Version 1.Advanced Security Services only standard IP access lists may be used.192 GMT: %FW-4-ALERT_ON: getting aggressive.0 Oct 10 20:01:30.2. Oct 10 20:02:19.1c 245 . you can only restrict scanning by source IP address.from 14. The example below shows how to apply our IDR1 rule for traffic coming into the 14. .1.10. count (0/400) current 1-min rate: 0 South# South# show ip audit statistics Signature audit statistics [process switch:fast switch] signature 2001 packets audited: [3123:6246] signature 2004 packets audited: [4:8] signature 3050 packets audited: [902:0] Interfaces configured for audit 1 Session creations since subsystem startup or last reset 1193 Current session counts (estab/half-open/terminating) [0:0:0] Maxever session counts (estab/half-open/terminating) [0:51:0] Last session created 00:01:50 Last statistic reset never Post Office is not enabled . a security audit host used to perform test scans).1.from 14. The example below shows how to disable two signatures related to common ICMP packet types.1.6 to 14. This session was captured during a small TCP port scan.0 Oct 10 20:01:15.696 GMT: %IDS-4-TCP_SYN_ATTACK_SIG: Sig:3050:Half-Open Syn Flood . This makes the facility most useful for reducing false positives caused by specific trusted hosts (e. with IDS South(config-if)# ip audit IDR1 in South(config-if)# end South# Detection Sample The transcript below shows the log messages that the IOS Firewall IDS generates and statistics it records.2. the router IDS was configured to send alarms to normal IOS logging rather than to an IDS Director.0 network.2.6 to 14. count (51/500) current 1-min rate: 501 .10. because only packets which have been permitted by any in-bound ACLs on other interfaces will be scanned. Note that. South(config)# interface eth0/0 South(config-if)# description External interface.10.1. South# Oct 10 20:01:13. in either the in-bound or outbound directions. It is also possible to disable and restrict particular IDS signatures.644 GMT: %FW-4-ALERT_OFF: calming down. For details. for this case.

that is. nmap. and/or a Cisco Secure Director.3. because internal network misuse is not uncommon and potentially as detrimental as an external penetration. the router console. Despite its speed and excellent location (no forwarded packet can avoid being scanned) the small signature database and inability to correlate different events prevent the IDS from being effective against many realistic attacks. Special Note: Because it is performed as part of packet routing. such as distributed scans. and possibly help you understand threats to your network more quickly. while the console is usually in a secure facility.1) The IOS Firewall IDS facility can help detect and track analysis of your network by remote parties. With common attacks stopped and logged. buffer overflows. Security Considerations for Using IOS Firewall IDS Before an intruder can successfully penetrate a network. it is not analyzed.5. you cannot use this facility to detect attacks from one host to another on the same LAN. Recommendations The Cisco IOS Intrusion Detection System does not provide comprehensive intrusion detection as a stand-alone feature. Many tools and technique exist to help attackers gain this information (e. one or more dedicated IDS should be deployed on internal networks to provide more comprehensive coverage and analysis.1c .Router Security Configuration Guide 5. 246 Version 1. The IDS can report intrusions to a given host's syslog. nor was it designed for this purpose. It is best used to supplement more complete intrusion detection packages. By default. both require a human to actively monitor them to provide real time information. it is difficult to monitor an attack against the network because both the syslog and the router console do not lend themselves to instant reporting .the syslog writes a textfile. which is critical in cases of intrusion. discussed in Section 6. and attempted root logins.4. Keep this in mind when planning intrusion detection for your network. Thus. The IOS IDS cannot stand alone as a complete network defense package. they must have information about it.g. Without the Director. This will provide simple ID at the edge of a protected network and stop simple attacks. Cisco IOS IDS cannot monitor internal traffic. if a packet does not need to be routed. This can be most efficiently accomplished by installing the IOS IDS at a border point with the firewall configured.

[8] “How to Configure IPSec Tunneling in Windows 2000”. 2004. O’Reilly Prentice-Hall. The master document for IPSec. A Technical Guide to IPSec Virtual Private Networks.6. includes extensive remarks about VPN architecture. Cisco Internetworking Solutions Guides... Cisco Product Support Documentation.” RFC 2401. plus and technical detail about IKE and VPN design. and related configuration commands. J.aspx?scid=kb. and Zwicky.0 Network Security. [3] Doraswamy. Intranets. 2004. Cisco Press. Cooper. and Virtual Private Networks. “Security Architecture for the Internet Protocol. available at: http://support.. Contains a good overview of IPSec. 252735. available at: http://www. D. 1999.Advanced Security Services 5. CBAC. D.en-us. S. 2001. R. including IPSec. Cisco Systems. Microsoft Corporation. Version 1.252735 Contains detailed information about setting up IPSec in Windows 2000. ps1018/prod_configuration_examples_list. Cisco Systems. S. 1998. ps2138/products_maintenance_guide_book09186a008007da16. [6] “Cisco IOS Firewall Configuration Examples and TechNotes”. IPSec: The New Security Standard for the Internet. [7] “Cisco Secure VPN Client Solutions Guide”. 2000.1c 247 .cisco. N. and Atkinson.html Detailed information on configuring Cisco VPN client [4] Kent. This highly technical book provides detailed explanations and pragmatic advice about IPSec. Auerbach Publications. Article No. Authoritative source for in-depth descriptions of security-related IOS facilities. Microsoft Knowledge Base. 1999.D. References [1] Chapman. A seminal reference for understanding firewalls and the principles for building them. [2] Cisco IOS 12. 2nd Edition. and Harkins. available at http://www. 2000. [5] Tiller. Building Internet Firewalls.html This page offers several CBAC documents and configuration examples.

[12] “Secure Shell Version 1 Support” software/ios121/121newft/121t/121t1/sshv1. [10] “An Introduction to IP Security (IPSec) Encryption”. D. and Silverman.pdf A detailed overview of the IOS Firewall IDS software/ios120/120newft/120t/120t5/iosfw2/ios_ids.pdf This tech note offers detailed information about Cisco IPSec support.htm A short overview of SSH features in IOS 12.1(1)T. Provides broad and detailed coverage of SSH Intrusion Detection.Router Security Configuration Guide [9] “Overview of Secure IP Communication with IPSec in Windows 2000”.com/ pcgi-bin/Support/browse/psp_view. available at: http://www. [13] “Cisco Security Advisory: Multiple SSH Vulnerabilities”. 1998. 231585.6. with examples.aspx?scid=kb. Cisco T. Microsoft Cisco release notes. 2000. software. SSH The Secure Shell – The Definitive Guide.en-us. available at http://www. available at: http://www. IOS 12. visit: http://www. including a list of the supported information and attack signatures. 2000.E. November Article No. A good introduction to intrusion detection concepts and techniques. Wiley. Cisco Systems.J. O’Reilly Associates. For more documents and examples about IPSec. Cisco Technical Notes. available at: http://support. [14] Barrett. Cisco Systems. available at http://www. available at: http://www.. May 2003. and usage. IOS 12.pdf Provides detailed information on using error messages and debug facilities to track down problems in IPSec configuration. Revision . [16] Escamilla.1c . [15] “Cisco IOS Firewall Intrusion Detection System”.231585 A good overview of IPSec features in Windows 2000. Cisco Technical Notes. 248 Version 1. R.1 release notes.html An overview of SSH vulnerabilities and IOS versions to which they apply. [11] “IP Security Troubleshooting − Understanding and Using debug Commands”. Microsoft Knowledge Base. Cisco Systems. May 2003.

1c 249 .Advanced Security Services Version 1.

Section 9. Packet sniffer programs are used to monitor traffic passing through the network and steal unencrypted passwords and SNMP community strings. this information can then be used to formulate specific attacks against the router. Testing Tools There are a variety of tools available for testing purposes. Testing and Security Validation 6. Scanners such as Fyodor’s nmap program can be used to scan for open TCP and UDP ports on a router interface.2.1c .3. Some of these services are enabled by default whereas other services are frequently enabled by users. testing should be performed at initial deployment of a router. Principles for Router Security Testing The border router is often the first line of defense when protecting against malicious network attack. 250 Version 1. Attack scripts are readily available on the Internet for numerous well-known exploits.1. and whenever major changes have been made to any part of the configuration of a router.Router Security Configuration Guide 6. Routers provide many services that can have severe security implications if improperly configured. 6. several denial of service (DOS) attacks and the newer distributed denial of service (DDoS) attacks have been highly successful against network devices. Additional tools are listed in the Tools Reference. Security testing provides a means of verifying that security functions are compatible with system operations and that they are configured in a secure manner. including some versions of IOS. Ideally.

but for IOS v. The default settings for these services depend upon the IOS version.3.11.2. there shouldn’t be any ports open.1. A serious known problem with Cisco IOS is that some default settings are not displayed as part of the router configuration listing.1.3. The examples below show nmap running under Linux.1. (Note: if IP unreachable messages have been disabled.3. DNS must be available for any operation referencing a host by name to succeed (e.3. To view the current operational configuration. use the EXEC mode command show running-config.1.250): Port State Protocol Service Version 1.12 by Fyodor (fyodor@dhp. For example. 2.11. for TCP FIN use -sF Starting nmap v. show the ‘udp-small-servers’ or the ‘tcp-smallservers’ in the configuration. the default is enabled. for Interesting ports on (14.2. Access lists should be tested thoroughly once assigned to an interface both to be certain that necessary traffic is permitted and unwanted traffic is denied.g. Telnet).250 –p 1-65535 Starting nmap v. run a port scan against the router. temporarily re-enable them before performing your UDP port scan by using the interface configuration command ip unreachable. as advised in Section 4. Additionally.2. To verify the entire configuration. the default is disabled.12 by Fyodor (fyodor@dhp. for IOS v.250): # nmap –sU -p 1-65535 14.1c 251 .2.) TCP Scan: The following command will perform a TCP scan against router North (IP address 14.Testing and Security Validation 6. 2. The nmap scanning program is a good tool for this purpose. Testing all allowed services will identify these dependencies.250): # nmap –sT 14.2. cross-check the ports that nmap reports open against the services that the router is supposed to be running.2. Interesting ports on (14. some services depend on other services in order to function.1. UDP Scan: The following command will perform a UDP scan against router North (14. Testing and Security Analysis Techniques 6.1.1. Functional Tests Functional testing provides assurance that the implemented configuration is the intended one.250): Port State Protocol Service If VTY (Telnet) access is not allowed. The above command would not.250 Warning: -sU is now UDP scan.

org/summaries/.. visit the CERT site: http://www. WARNING: RUNNING ATTACK SCRIPTS AGAINST AN OPERATIONAL ROUTER MAY DEGRADE ROUTER PERFORMANCE. This attack has at least two victims – a target system and one or more reflector systems. Physical security of the router should provide protection from close-in (non-network) access. One popular DoS attack is the ‘smurf’ attack. including recent developments and links to specific DoS advisories. For a more detailed discussion of Cisco router access security and remote administration. 252 Version 1. DoS attacks are typically based on high-bandwidth packet floods or other repetitive packet streams. The attacker sends a continuous stream of ICMP echo requests (‘pings’) to the broadcast address of a reflector subnet. OR EVEN CAUSE THE ROUTER TO CRASH! If the filters are improperly configured. Attack Tests Attack testing can provide some assessment of the router’s robustness.cert. visit: http://www. The source address in these packets is falsified to be the address of the ultimate target. i. remote access must be limited using authenticated logins or.cert.html. flooding the target and wasting bandwidth for both victims. Each packet generates a response from all hosts on the reflector subnet. Some of the most common attacks on the internet are denial of service (DoS) attacks. The router should either refuse the request or prompt for a password. The reflector networks receiving these echo requests can block the attack at their routers by using the interface configuration command no ip directed-broadcast (see Section 4. For a detailed discussion of the smurf attack.2.Router Security Configuration Guide 6. make sure that all attack testing is coordinated with those responsible for the network and choose a test time when the network usage is likely to be low. DO NOT perform attack testing against any network until you have received organizational and legal approval to do so. The easy availability and effectiveness of DoS scripts on the internet make these attacks a favorite among hackers. Connecting to an outside network exposes the internal network and the perimeter router to many potential risks.e.1. if possible. the network is still at risk of attack. Perform testing in a lab or testbed environment before testing in the operational environment. and the Cisco whitepaper “Improving Security on Cisco Routers” [1].1c . DO NOT perform attack testing against an operational router without first considering the possible consequences and having a recovery plan. or not applied to the interface.2). For a general overview of DoS. On the network.3. particularly those without the skill to create their own tools. how the router will perform under the stress of an For more information on the effects of DoS attacks. One of the most important security concerns is access to the router itself. Once access to the router has been secured. some of these attack tests can have the same effect as a “real” attack from a malicious source. To test the remote availability. telnet to the router. remote logins should be disabled. consult Section 4. When you do perform testing on the operational network. read Craig Huegen’s paper [9].

More information about DDoS attacks is available from references [3]. This feature checks each packet arriving at the router. Cisco routers can help prevent the system behind the router from being an unwitting participant in a DDoS attack by using the ip verify unicast reverse-path interface command (Section 4. requiring the attacker to send more phony packets to flood the service. the timer is reset and the connection is established. i.e. For Cisco routers running IOS v. described below.1c 253 . Unlike a “typical” smurf attack. and it is only available in IOS v12. the requesting client sends a SYN packet to the server. The server responds with a SYN/ACK packet.11. In a SYN flood. packets arriving at the router Version 1. Cisco provides a paper titled “Defining Strategies to Protect Against TCP SYN Denial of Service Attacks” [4]. reserved addresses (defined in RFC1918).0. Asymmetric routing will not work with this feature. The host could increase the size of the connection queue. the SYN flood..0).2 or 11. Trin00. Blocking access to the service under attack is usually not feasible and would accomplish precisely what the attacker set out to do. and [8].4.0. victims of a SYN flood do have some options. tracking the attacker is extremely difficult. Also. which uses a limited number of reflector systems. Additionally. The real attacker is able to amplify the DoS flooding while being removed from the attacking machines. adds an entry in the connection queue and starts a timer. The victim responds with a SYN/ACK to the unreachable host and waits for the ACK. In a normal TCP connection request. The attacker continues to send these forged SYN packets at a rapid rate until the victim’s connection queue is filled with half-open requests. takes advantage of the TCP three-way handshake procedure to deny service to the victim. These distributed denial of service tools (DDoS) pose a major threat to networked systems and have the potential to severely impact normal business activities. filters can be used to prevent IP spoofing in a manner similar to the ip verify unicast reverse-path feature discussed above. The requester then completes the handshake with an ACK packet. these tools employ many compromised systems to simultaneously attack a single target. The effect of this attack is to deny TCP services such as e-mail. For more options. Tribal Flood Network 2000 (TFN2K) and Stacheldraht. file transfer or web traffic to legitimate users. four historically popular ones are called Tribal Flood Network (TFN). [5]. [4]. changing the source IP address to hide the true source of the packet. arriving packets should not have source addresses of all 0’s or all 1’s or the loopback address (127.3. An integral part of DoS and DDoS attacks is IP spoofing. similar protection can be achieved by filtering for IP spoofing. Access lists should check that no packets arriving from the outside network contain a source address of either the internal network or the well-known. the packet is dropped. Another common DoS attack. if the source IP address does not have a route in the CEF tables pointing back to the same interface on which the packet arrived. non-routable. thus emptying the queue of half-open connections more quickly. an attacker sends a TCP connection request (SYN) packet with an unreachable. The host could also decrease the wait time for completion of the three-way handshake. However. the ACK doesn’t arrive and the TCP handshake never completes. the queue entry is removed.0. spoofed source address to an open port on the target.7).Testing and Security Validation Enhanced denial of service tools have recently become available on the Internet. There are many such tools in circulation.

255 0. For an in-depth discussion of TCP flooding and IP spoofing.2. 11. There is a Cisco syslog vulnerability that may cause the IOS software to crash if an invalid user datagram protocol (UDP) packet is received on the syslog port (port 514). This input access list must block traffic destined for UDP port 514 at any of the Cisco IOS device’s own IP addresses.0 eq 514 ! Deny network-specific broadcasts to the 14. This problem can be prevented by applying the appropriate input access list to all interfaces that might receive these packets.0. as well as at any broadcast or multicast addresses on which the device may be listening. and early (non-GD) releases of 12. The router should drop these packets at the perimeter and log the events. To verify the anti-spoofing configuration. By sending such packets continuously.0.0.0. consult [7]. At least one commonly available vulnerability scanner can generate these UDP packets. send packets to the router’s internal interface with source addresses that are not legitimate internal addresses. The following example shows an access list to block the port 514 UDP traffic. The packets should test the ability of the router to detect both internal addresses and reserved addresses that should not arrive at an external port.2. (Note that the internal network may be using one of the RFC1918 reserved addresses with NAT performed at the router.Router Security Configuration Guide from the internal network should not have a source address that is not one of the legitimate internal addresses.0. the access list on the internal interface will recognize such an address as legitimate. send a series of packets with modified source addresses to the external interface. an alternative fix would be to deny all syslog traffic arriving on that interface.0 eq 514 ! Deny packets addressed to router interface access-list 101 deny udp any host 14.2.0.) This check will prevent the internal network from being used as a launch point for a source IP spoofing attack. The goal here is to catch packets with a source address of an external network or a reserved address that is not being used by the internal network. To verify outbound anti-spoofing.255 eq 514 ! Deny old-style unspecified net broadcasts access-list 101 deny udp any host 0.0 net access-list 101 deny udp any 14.1c .0 31.0 eq 514 access-list 101 deny udp any 14.3AA. an attacker might be able to completely disable a Cisco IOS device until the affected device is reconfigured to drop the attack traffic.20 eq 514 ! Apply to input interface of router North interface eth0 ip access-group 101 in 254 Version 1. RFC 2267 discusses network ingress filtering and defeating DoS attacks which employ IP source address spoofing.0. If a specific interface is not expected to forward legitimate syslog traffic.0.255. it might require an administrator to physically visit the attacked device to restore service. the router should again drop the packets and log the events. Some vulnerable IOS devices will “hang” and must be manually restarted by reset or power cycle.0. ! Deny all multicasts and all unspecified broadcasts to port 514 access-list 101 deny udp any 224. This vulnerability affects unpatched versions of IOS 11.0.0. in this case.2.0 0.

the decision is made to go forward with this testing. Other sites for exploit information and code are listed at the end of this section.3. For more information. the attack scripts are readily available from many sources on the internet. WARNING: RUNNING AUTOMATED ATTACK TOOLS ENTAILS SIGNIFICANT RISK! It is easy to accidentally auto-scan more systems than you intended. visit: http://www.3.php. Packetstorm Security has several DoS exploits. web servers.01) now contains tests to find hosts infected by DDoS agents. CyberCop generates reports from scan results that include information about detected vulnerabilities: a description of the vulnerability.asp. The Security Administrator’s Integrated Network Tool (SAINT) and the Security Administrator Tool for Analyzing Networks (SATAN) are publicly available tools. 6. Version 1.1c 255 . Internet Scanner analyzes the scan data and provides reports containing vulnerabilities identified along with recommended corrective actions. level of risk and suggestions for fixing/mitigating the The latest version of Internet Scanner (6. planning and coordination. Exercise caution when using tools like CyberCop. visit the IIS web site. CyberCop offers monthly updates consisting of new modules and security hotfixes for new and evolving vulnerabilities. or to touch systems for which you have no legal authority. available under http://packetstormsecurity. the router could crash or hang! As mentioned above.pgp. However. Internet Scanner includes nearly 600 total tests. Internet Scanner probes the network’s communication services. assessment page: http://www. and monitor the tools closely while they are operating. At the time of this writing. and updates containing the latest tests and security checks are available for download daily. key applications and routers for those vulnerabilities frequently used by malicious users to investigate and attack security concerns. Internet Scanner is also a network vulnerability analysis and risk assessment product. if the router is running a vulnerable version of the IOS software and the access list is not properly configured or not applied. operating systems. Mechanisms for Automated Testing There are a number of products available to automate the testing running DoS attack scripts against the router can have very serious and undesirable consequences. SAINT. CyberCop Scanner performs comprehensive evaluations of and http://packetstormsecurity. CyberCop Scanner from Network Associates and Internet Scanner from ISS are two popular commercial products. If. For more information.Testing and Security Validation This vulnerability can be tested by sending a UDP packet to the router’s port 514. or SATAN. after careful consideration. always double-check the addresses to be scanned. firewalls and screening routers by scanning them and performing extensive tests to identify known vulnerabilities.

An analysis of the packets arriving on the serial interface can identify the specific attack being used. NIS. ftp. In a typical smurf attack. The initial data collection can then be used to investigate any potential security problems. Assume the following access list is applied to interface 14. Detecting Attacks As mentioned in section 6.porcupine.2. use the command show accesslist 102.saintcorporation. consult Section 4. To download a copy of SATAN. SATAN offers a tutorial that explains the problem and the potential impact. visit: http://www. To see the number of matches for each line in the access list. SATAN was designed to help system administrators responsible for the security posture of their this feature exposes the real security implications inherent in network trust and services.4. these are the addresses of the reflector sites. 256 Version 1. The fourth line therefore matches only packets that are connection requests. The third and fourth lines of access list 102 characterize TCP traffic. SATAN also provides corrective actions including configuration changes.Router Security Configuration Guide SAINT gathers information about remote hosts and networks by examining network services such as finger. or possibly disabling services. In normal operations.3. tftp. it is a tool for investigating the vulnerabilities of remote systems. If the incoming traffic consists mostly of ICMP echo requests.20 of router North: access-list 102 permit access-list 102 permit access-list 102 permit access-list 102 permit access-list 102 permit interface serial 0 ip access-group 102 in icmp any any echo log-input icmp any any echo-reply log-input tcp any any established tcp any any log-input ip any any This access list does not filter out any traffic but does separate the traffic by types.3. the TCP SYN packets. visit ftp://ftp. For each problem found.1c . IOS access lists can be used to characterize the different packet types and to tentatively identify DoS attacks. the attack is probably a smurf attack where North is a reflector. SAINT can also be configured to examine trust and dependency relationships in the target network.0. including a FAQ. The keyword established in the third line matches any TCP traffic with the ACK bit set. that is. 6. installing vendor hotfixes. NFS. denial of service attacks are very common on the internet. The signature of a smurf attack where router North is the ultimate target would show most of the packets as ICMP echo replies. For more information. rsh commands and other services. any TCP traffic that is not a connection request. a necessary first step in countering DoS attacks. the source addresses in the echo reply packets are limited to a few networks.3. The vulnerabilities discovered are then reported to the user without actually exploiting For more information about access lists. SATAN systematically proceeds through a target network probing for common networking-related weaknesses and security problems. a tutorial and the latest version of SAINT.2 above.

SYN floods usually contain packets with invalid source addresses. An on-going problem is that new attacks can appear so fast on the internet that countermeasures are not immediately available. the upstream provider to the victim may be able to filter out all ICMP echo replies to the target network.3. It is almost impossible to protect a network from denial of service attacks.3). these SYN packets typically outnumber other TCP packets many times over. In a distributed attack.1c 257 . or not sufficient to stop the attack. if it can be determined that the originators of the attack are limited to a few addresses. logging such traffic (as recommended in Section 4. This paper gives an overview of denial of service attacks and a detailed discussion of using access lists to categorize packets. 6. it may be possible to apply specific filters at the external interface of the border router.3) will let the administrator determine if such source addresses are present. Second. the ultimate target cannot filter out the attacking addresses. both inbound and outbound. The best advice is to configure the router to check for IP spoofing. Version 1.cert. In this case. In a SYN and implementing the latest patches from the vendors.Testing and Security Validation TCP SYN packets account for a third or less of the total TCP traffic. the only response may be to contact the reflector sites to reconfigure their networks to shut down the attack. Attack Reaction Options It is difficult for the ultimate target of denial of service attacks to stop or even blunt an active attack. Still.5. if filtering is not possible. There is a paper available on the Cisco web site titled “Characterizing and Tracing Packet Floods Using Cisco Routers”. First. and to only allow services that are needed (see Sections 4. the only defense is to be vigilant about security and to keep up with that latest security news by regularly checking a site such as CERT (www. The paper also describes how to trace DoS attacks and the complications inherent in packet tracing [2]. this filter should only be in place temporarily and only as a stopgap measure.2 and 4. Also.

Second.g. Once you have RAT installed. routerconf-1. This option involves supplying an administrator login name. RAT is written in Perl.cisecurity. If you supply the address of the router. routerconf-1. routerconf-1. The Center for Internet Security (CIS) maintains RAT. convenient native binaries of RAT are available. (e. The rat program itself produces four output files for each router configuration tested. RAT uses its own utility called ‘snarf’ to do this.g. (e.g. On Windows platforms. consult the RAT documentation. Using this approach.1 or later. rat has the ability to log in to the router via Telnet and obtain the configuration directly.ncat_out.txt) • An HTML version of the report.html) 258 Version 1. time and logging configuration. and a username and passwords.g.4. • An ASCII file that lists all the raw results for the configuration in a compact text and select the “Cisco IOS Router” link. although ActiveState’s ActivePerl may also be used. For more information. First. password. and requires a Perl installation to run on Unix or Linux. consult the download notes at the CIS web site for details. The default rules that RAT applies for Cisco IOS routers are in close agreement with the material presented in Section 4 of this guide.6. the procedure for running it is fairly simple. access lists. and it provides a means for an administrator to quickly check whether their router meets an IOS security ‘benchmark’ designed by a panel of industry and government security experts. and router enable password on the RAT command line. The rat program can take the configuration input in two ways. providing the IOS configuration of the router as input. The procedure for installing RAT varies between platforms. its interfaces. the current version of RAT was 2. It will run on any Unix or Linux platform supporting Perl 5. The RAT program can also accept the router configuration as a text file. Ncat_config will ask you a series of questions about the router’s role. RAT is not difficult to use.txt) • A formatted text version of the report. Using the Router Audit Tool The Router Audit Tool (RAT) tests whether a Cisco IOS router configuration complies with a set of community consensus security rules.2P. services. you must run the command ncat_config to customize the rules to your particular router(s). which should only be done on a completely trusted host and network. or via FTP) and place the files into the directory where the rat command will execute.ncat_report.Router Security Configuration Guide 6.. As of September 2005. To download RAT visit the Internet web site www. (e.1c . and sponsors the working group that defines the default RAT rules. run the rat command itself. you gather one or more router configurations by other means (e. and generates HTML reports detailing which rules the configuration passes and fails. via the console port on each router.

html.conf/ Checking: northconf-2aug03 done checking northconf-2aug03.html (cisco-ios-benchmark.1c 259 .conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-1.txt) RAT Example The transcript below shows a small RAT session (version 2..html.conf ncat_config: c:\CIS\RAT/etc/configs/cisco-ios/cis-level-2.ncat_fix. D:\routeradmin> set PATH=c:\CIS\RAT\bin. D:\routeradmin> Note that this example does not show all of the questions posed by ncat_config.Testing and Security Validation • A text file that contains IOS command to fix the problems identified in the report.html).ncat_fix.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-2. .txt.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-1. ncat_report: writing all.ncat_report.ncat_report. ncat_config: Check rules and data related to data flow [Yes] ! ncat_config: Apply standard routing protections [Yes] ? yes ncat_config: Forbid directed broadcasts (on IOS 11) [yes] ! ncat_config: Forbid directed broadcasts (on IOS 12) [yes] ! ncat_config: Forbid IP source routing [yes] ! ncat_config: Apply some or all of CIS Level 2 rules [No] ? no D:\routeradmin> rat northconf-2aug02 auditing northconf-2aug03.%PATH% D:\routeradmin> ncat_config ncat_config: Select configuration type [cisco-ios] ? cisco-ios ncat_config: Applying rules from: ncat_config: c:\CIS\RAT/etc/configs/cisco-ios/common. Version 1.conf ncat_config: Apply some or all of the rules that are selectable [Yes]! ncat_config: Apply some or all of CIS level 1 rules [Yes] ? yes ncat_config: Check rules and data related to system management [Yes] ! ncat_config: Use local authentication [Yes] ? yes . ncat_report: writing all. ncat_report: writing rules. routerconf-1.. Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/common.txt. (e. ncat_report: writing northconf-2aug03. Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/common.conf ncat_config: c:\CIS\RAT/etc/configs/cisco-ios/cis-level-1. ncat_report: writing northconf-2aug03. ncat_report: writing all.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-2.g.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/local.txt.1) run on Windows. .ncat_fix.txt.conf/ ncat_report: writing northconf-2aug03.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/local.

Examine each rule failure that RAT reports.1c . your router must meet your local security policy. 260 Version 1. RAT is simply a useful tool to help you find problems and areas for improvement. treat it as a potential issue. do not follow them blindly.Router Security Configuration Guide RAT Caveats Use the RAT benchmark reports as an aid in securing your routers. In the end. and check the router settings related to the rule carefully.

Securify Cisco Systems. 2002. 2000. CERT Coordination Center. 1999. Cisco Technical Tips. a web-based tool that registered Cisco customers can use for a cursory security analysis of their IOS configurations. Packet Storm. 2000. Cisco Systems. Cisco Systems.html A good overview of DoS attack 2000.Testing and Security Validation 6. available at: [9] Huegens. available at: http://www. CERT Incident Note IN-99-07. Cisco White Papers. CERT Coordination Center.html [4] “Defining Strategies to Protect Against TCP SYN Denial of Service Attacks”.org/archive/t1497. available at: http://packetstormsecurity. CERT Advisory CA96. also includes a reference to the Cisco IOS Output available at: http://www. [6] “Distributed Denial of Service Tools”. C. Software Engineering Institute. 1997. available at: http://www. available at: http://www.html A good summary of basic IOS router security practices.cert.html Version 1999.html Detailed guidance on tracing certain kinds of DoS attacks. Cisco White Papers. available at: http://www. References Web Sites and On-Line Resources [1] “Improving Security on Cisco Routers”. Cisco Technical Tips.governmentsecurity.html [7] “Topic: TCP SYN Flooding and IP Spoofing Attacks”.com/warp/public/707/21.1c 261 . Software Engineering Institute. available at: http://www. available at: http://www. [3] “Strategies to Protect Against Distributed Denial of Service (DDoS) Attacks”.21.cert. Cisco Systems. [5] “Denial of Service Attacks”.html [8] “Distributed Attack Tools”. “The Latest in Denial of Service Attacks: Smurfing”. CERT Coordination Center. Software Engineering Institute. [2] “Characterizing and Tracing Packet Floods Using Cisco Routers”.

org/ 262 Version 1.http://www. “Exploiting Cisco Routers (Part One)”.php SAINT ISS Internet Scanner Security Configuration Guide [10] http://www.html SATAN .nl/exploits/DoS Automated security scanning and testing tool sites: RAT http://www.saintcorporation. .cisecurity.securityfocus.nessus.http://www. available from: Additional Exploit-Related Pages: NESSUS . Additional General Exploit Information Sites: http://www.porcupine.iss.

• Protecting physical access to the switch itself • Controlling virtual access to the switch. routers can supply critical protection at network boundaries by filtering traffic (see Section 4. Note that some switches can act as routers between their separate VLANs. Additional Issues in Router Security This section describes a few areas of network technology that will probably have an effect on router and network security in the near future. The security concerns for switched networks and switches correspond directly to those for routed networks and routers. Many Ethernet switches have the ability to maintain one or more separate virtual LANs over the same physical cables and switches. while others require a real router. Routing and Switching As network bandwidth demands continue to increase. IP routing will increasingly be replaced by layer 2 switching in high-performance applications. the topics described below are merely a select few of the many technologies that network security administrators will have to incorporate into their security plans and policies in the next few years. The list is not comprehensive. in a network environment that is predominantly based on switching.1. 7. One feature of switched environments that might be usable to improve security is Virtual LAN switching.Additional Issues in Router Security 7. Therefore. while removing the ability to impose some security restrictions. Switches typically have limited or negligible filtering capabilities. For example. each individual host and device must be configured securely rather than relying on protection at their LAN boundaries. Version 1.3). The diagram below shows how virtual LANs can be set up to emulate two physical LANs spread across two switches.1c 263 . including user authentication and authorization • Updating the operating system when necessary to fix known vulnerabilities • Preventing unauthorized modification of the switch configuration • Disabling unneeded services and features Switching imposes new risks.

6) Virtual View Host B (14.2.0/24 Router LAN 2 14.2.1c .2.3.Router Security Configuration Guide Internet Host A (14.4.3) Host C (14.0/24 Host Y (14. 264 Version 1.2) Host X ( Host Z (14. but it is clear that VLAN security will grow in importance in the next few years.3) LAN 1 14.6) Host Y (14.2.7) Host B ( Host X ( Figure 7-1: Virtual LAN Switching More investigation is needed to determine the security roles and policies for configuring VLANs. Host C ( Host Z (14.8) Internet Real Construction Router LAN Switch 1 LAN Switch 2 Host A (14.

and other protocols 576 bytes optional. In the mid1990s. and later. 8-30 bits 20-60 bytes limited number of small IP options sender or any intermediate router allowed to fragment mixture of non-IP (ARP). hundreds of IPv6 address blocks had already been allocated to network carriers and enterprises all over the world. IP version 6 (IPv6) was first standardized in 1995.1c 265 . The US government and US DOD have set ambitious goals to complete substantial migration by 2008. due use of transition mechanisms and interactions between IPv4 and IPv6 features. and should be published by mid-2006. Version 1.2T. it became clear that the Internet would eventually outgrow the address space and capabilities of IPv4. The Internet community is gradually migrating from IPv4 to v6. ICMP. Defining best practices and configuration guidelines for routers handling IPv4 and IPv6 may be complicated. an IPv6 supplement to this guide is in preparation. Security best practices for IPv6 are not yet well-established. For more information. The table below gives a few comparisons between IPv4 and IPv6. and the IETF initiated efforts to standardize a successor.Additional Issues in Router Security 7.3. Property IPv4 IPv6 Address size Network mask size Packet header size Packet-level extensions Fragmentation Control protocols Minimum allowed MTU Path MTU discovery 32 bits varies. although some recommendations exist (see [15]). and the standard revised to its current form in 1998. During the transition period. such as [16]. 12.2. consult any good book on IPv6.nsa. not widely used 128 bits usually fixed. The supplement will be available from www. By mid-2005. 64 bits 40 bytes (fixed) unlimited number of IPv6 extension headers only sender may fragment all control protocols based on ICMPv6 1280 bytes mandatory Cisco IOS routers support IPv6 in IOS versions 12. it is expected that many routers will have to support both IPv4 and IPv6 (so-called “dual stack” operation) as well as supporting transition mechanisms like encapsulation and At the time of this writing. IPv6 The Internet was built using version 4 of the Internet protocol (IPv4).

It offers a great deal of flexibility.g. MPOA allows routers and other network devices to take advantages of advanced ATM facilities (like ATM quality-of-service). The security implications of choosing one of these modes over another are not yet entirely clear. and offers very simple configuration at the cost of flexibility and performance. ATM and IP Routing Asynchronous transfer mode (ATM) switched networks are popular for backbones and long-haul high-speed network links. Classical IP – This is the oldest technique. LANE – LAN Emulation (LANE) is a fairly general. most of which is out of the scope of this guide.1c . 2. 266 Version 1. MPOA – Multi-Protocol Over ATM (MPOA) is a highly flexible set of mechanisms for transporting IP and other protocols over ATM switched networks. but requires a great deal of configuration to deploy. 1. Cisco routers support three mechanisms for sending IP traffic over ATM switched networks. Used with LANE. Sometimes. Frame Relay). ATM is a very big topic.3. 3.Router Security Configuration Guide 7. standardized technique for extending an IP LAN over an ATM switched network. the boundary between switched ATM and routed IP will be a switch or router with one or more ATM interfaces and one or more traditional LAN or WAN interfaces (e. Ethernet.

In an MPLS network. Each label switch router (LSR) makes forwarding decisions based only on the contents of the label. Label switching allows routers to make forwarding decisions based on the contents of a simple label. MPLS is defined and specified by the IETF.4. MPLS uses label switch technology to simplify routing and enhance overall network performance. Packets are forwarded along a label switch path (LSP). At each hop. MPLS enhances the services that can be provided by IP networks. offering traffic engineering (TE) . Version 1. the LSR removes the existing label and applies a new label which tells the next hop how to forward the packet. rather than by performing a route table lookup based on destination IP address.1c 267 . Multi-Protocol Label Switching (MPLS) MPLS is an emerging high speed switching protocol typically deployed in the network core of a large enterprise such as an ISP. Quality of Service (QOS). incoming packets are assigned a label by a label edge router (LER). and Virtual Private Networks (VPN) capabilities.Additional Issues in Router Security 7. the current version is described in RFC 2547 and RFC 3031.

a router or any other device must possess a copy of the correct root and authority certificates.2 probably will not scale well enough to satisfy users’ needs. employing public keys and public key certificates to set up IPSec-protected links on the fly Security configuration issues are likely to be important in deployment of large dynamic VPNs are listed below. the simple configurations described in Section 5. IPSec and Dynamic Virtual Private Networks Section 5. 268 Version 1. but not for authorization or access control. As deployment of VPNs becomes more common. removing certified principals from the trusted community is very important. Installing certificates and setting up authorities on Cisco routers is complex but well-documented. consult Cisco IOS 12.509 Certificate Revocation Lists (CRLs) and the Online Certificate Status Protocol (OCSP).1c . PKI standards define various data formats and protocols for defining revocations and for checking certification status. For complete information on the IPSec and dynamic VPN capabilities of Cisco IOS 12. • PKI enrollment and obtaining certificates – To participate in a dynamic VPN based on Public Key Infrastructure (PKI). Currently. It may be necessary to configure revocation checking on routers participating in dynamic VPNs. VPNs will need to be dynamic. and it must have its own certified public key and private key.2 explains some of the basic features of IPSec. support the ability to protect certain traffic based on its protocol and port numbers. However. IPSec and Virtual Private Network (VPN) configuration are complex topics. • Cryptographic issues – Selection of uniform key sizes and cryptographic algorithms will be a contentious issue in VPN deployment. • Designating traffic to be encrypted – Cisco routers.0 Network Security [2]. including X. depending on the installed IOS version and feature set. and most other VPN systems.0. There are also trust issues in any large VPN deployment: are all members of the VPN trusted equally? In general. there are no uniform guidelines for selecting traffic to protect. IPSec is most useful for integrity and confidentiality assurance.Router Security Configuration Guide 7.5. • Certificate revocation – In any large-scale PKI. Cisco routers currently support only a small complement of algorithms. To achieve scalability.

0/24 Remote User 14.2.6. over IP from a remote router or network access server to a central router.1. Virtual Private Dialup Networking Cisco routers support tunneling dial-up protocols. and it is illustrated in the figure below. over IP networks. VPNs.2. from a remote part of the network. Version 1. like PPP. the primary such protocol used today is called the Layer 2 Tunneling Protocol (L2TP).4. the security for a VPDN service depends on use of IPSec between the two ends of the tunnel: the remote network access server and the central router. but it seems possible that small deployments could use static IPSec tunnels as described in Section 5. such as Ethernet or PPP. Virtual Private Dial-up Networking configured between these two devices Access Server 126. There are several approaches to doing this.2.1c 269 . each of which has different security issues.9.185 Figure 7-2: Overview of Virtual Private Dial-up Networking In general. and actually use their home base LAN address. This kind of tunneling architecture is called Virtual Private Dial-up Networking (VPDN). This is an area that needs further study.6.9.19. 7.9.Additional Issues in Router Security 7. Tunneling Protocols and Virtual Network Applications As VPNs become more popular and widespread. and protected networks from remote sites.10 modem Virtual Connection Inside LAN 14. Standard protocols exist for tunneling layer 2 protocols. expect a corresponding increase in mobile users and addressable devices expecting to join home base networks. Use of such tunneling protocols allows remote users to join a LAN.2.29 Internet home base router dial-up File Server 14.

Cisco IOS 12. users will begin to demand quality-of-service assurances.Router Security Configuration Guide 7.1 and later support centralized application of RSVP policies. and configuring it securely will be challenging. but the security issues associated with this facility have not yet been explored. IP Quality of Service (QoS) and RSVP The Resource reSerVation Protocol (RSVP) is the Internet standard protocol for setting up Quality-of-Service (QoS) parameters for traffic in routed IP networks. but the security issues involved in that integration have not been explored.7. consult Section 4. by providing mechanisms for denying bandwidth to authorized users. Extensive guidance already exists for integrating IP QoS (RSVP) with ATM QoS. RSVP configuration will probably be a contentious issue. While the RSVP protocol itself includes provisions for authentication and authorization.0 and later support RSVP and QoS features. In general. Quality-of-service support offers the potential for substantial denial-of-service attacks. 270 Version 1. such as IP video-conferencing systems. For more information about access lists. begin to gain wide acceptance. On routers that support RSVP but that do not need to provide any QoS guarantees. key management and deployment issues for RSVP security have not been resolved. Also.1c . Many releases of Cisco IOS 12. As bandwidth-hungry network clients. all RSVP messages should be denied on external interfaces using IP access-lists.3.

Version 1. It could then perform the security checks on remote DNS requests.Additional Issues in Router Security 7.250). Bind. The local secure DNS server would have to be configured to perform secure DNS requests on behalf of its non-security capable clients like the router. the domain name resolver in Cisco IOS cannot recognize or check DNS security extensions). it seems unlikely that it will become ubiquitous.1c 271 . it would be possible to gain some of the security benefits by configuring the router name server (configuration command ip name-server) to be a local secure DNS server. Cisco routers do not yet support acting as a secure DNS client (in other words. Secure DNS The Domain Name System (DNS) used on the Internet and other major networks provides the mapping between names (like central.8. In a network with secure DNS. Given the negligible deployment that secure DNS has enjoyed in the first couple of years that it has been widely to IP addresses (like 14.9. These security services are supported by the latest versions of the primary Internet DNS server implementation. and pass along only validated results. the mapping between names and addresses is fully authenticated and integrity assured. In a network with secure DNS. The basic DNS protocol offers no authentication or integrity assurance. The DNS Security Extensions standard defines comprehensive integrity and authentication facilities for DNS.

Router Security Configuration Guide 7.” RFC 2401. and Halpern. 1999. The updated standard for secure DNS. Upper Saddle River.9.” RFC 2535. R. Indianapolis. Cisco Press. References [1] Sacket. Lindell. Cisco Router Handbook. G. “Resource reSerVation Protocol (RSVP) – Version 1 Functional Specification. Berson. and Atkinson. The master document for IPSec.. New York. and Talwar. also good background reading for understanding the issues of hosting IP over ATM. 2000. [5] Kent. IN.” RFC 2205. IPSec: The New Security Standard for the Internet. [4] Doraswamy. 1998.. RFC 2747. [2] Cisco IOS 12. NY. Contains a good overview and substantial technical detail about IPSec and related topics. 1999. 2000. [3] Cisco IOS 12. “Domain Name System Security Extensions. “Classical IP and ARP over ATM”. RFC 2225.0 Network Security. S. The definition of Classical IP over ATM. “RSVP Cryptographic Authentication”. 272 Version 1. including IPSec and related configuration commands. Cisco Press. H. McGraw-Hill. “Security Architecture for the Internet Protocol. 1999.C. M. Describes the message authentication service to be used with RSVP. and Jamin. includes extensive remarks about VPN architecture. Authoritative source for in-depth descriptions of security-related IOS facilities. Z. defines the protocol structure and intent. This documentation volume includes extensive configuration information for Cisco ATM switching and LANE. [9] Laubach. The basic standard for RSVP.1c . [7] Braden. 1997. and Harkins.0 Switching Services. Contains a good overview of Cisco ATM facilities. Indianapolis. J. D. Prentice-Hall. 1998. 1999. and Virtual Private Networks. includes extensive discussion and examples. [8] Baker. IN. D. N. Intranets. NJ. [6] Eastlake.

and comparison with risks posed by IPv4.Additional Issues in Router Security [10] ito_doc/mpls_tsw.. D. P. Prentice-Hall. available at: http://www. PPP and L2TP.. Definition of the Internet standard tunneling protocol. 2003. including some coverage of security options. documents/v6-v4-threats. A very detailed overview of remote access and layer 2 tunneling.pdf A seminal analysis of network risks posed by IPv6. 2000. Cisco Press. [11] MPLS and VPN Architectures: A practical guide to understanding. A highly detailed guide to setting up MPLS networks. It also includes some security guidance for configuring Cisco routers. Jim and Pepelnjak. Addison-Wesley. with some analysis of its security. Provides detailed instructions for configuring IPv6 functionality on Cisco routers Version 1. 2001. V.1c 273 . including discussion of the relationships of IP. version “Layer Two Tunneling Protocol (L2TP). L2TP Implementation and Operation. Cisco Systems. [16] Desmeules.0. March 2004. [12] Shea. R. written from a network-oriented viewpoint. and L2TP. Cisco Press.. 2000. “IPv6 and IPv4 Threat Comparison and BestPractice Evaluation”. [14] Guichard. 2002. designing and deploying MPLS and MPLS-enable VPNs. Cisco Self-Study: Implementing Cisco IPv6 Networks. PPP. available at: http://www.. [15] Convery. P. Rubens. A very detailed overview of IPv6. U. “MPLS/Tag Switching”.pdf A short paper offering an overview of MPLS and a comparison with traditional routing. Ivan. [13] Cisco System. S. An in-depth treatment of L2TP itself. R.” RFC 2661. Internetworking Technologies Handbook. and Miller. 1999.

then turn off clearly unneeded facilities and services. chargen. To leave CDP running but disable it for certain network connections. [Section 4.Router Security Configuration Guide 8. discard. chargen) and TCP (echo.1] Implement access list filters by permitting only those protocols and services that the network users really need. Some services that should almost always be turned off are listed below. for more details consult the corresponding parts of Section 4. [Sections 4. CDP sends packets from your router once a minute or so identifying your router. and save memory and processor slots too. 8. Keep the offline copy of each router configuration in sync with the actual configuration running on the router. [Section 4. Start by running the show proc command on the router.3] Specific Recommendations 1. General Recommendations Comment and organize offline editions of each router configuration file! This sounds fluffy despite being a big security win. daytime) based services. Use the commands no service udp-small-servers and no service tcp-small-servers to turn these off.Cisco Discovery Protocol is used almost exclusively by Cisco RMON.5. and keep all it and all old versions under configuration management. [Section 4. discard.2] • Small services .miscellaneous UDP (echo.1c .1. apply the command no cdp enable to the appropriate interfaces. Shut down unneeded services . One of these is the UDP echo which is used in the ‘fraggle’ attack. The descriptions here are terse. Trying to deny just the ‘bad things’ is a losing proposition.3] Run the latest available General Deployment (GD) IOS version. 8. Appendices The sections below offer ancillary material and supplemental guidance for network and security administrators. along with some important general principles for maintaining good router security. Top Ways to Quickly Improve the Security of a Cisco Router This appendix describes the most important and effective ways to tighten the security of a Cisco router. Use the no cdp run command to kill the process and disable CDP globally.things that aren’t running can’t break. • CDP .5. [Section 4. This is invaluable for diagnosing suspected attacks or problems and recovering from them. References to appropriate parts of Section 4 appear at the end of each recommendation. and explicitly denying everything else.2] 274 Version 1.

2] Central(config)# interface eth 0/0 Central(config-if)# no ip directed-broadcast 3. Always start an access-list definition with the command no accesslist nnn to make sure it starts out clean. [Section 4.the Network Time Protocol. 4. the attacks will appear to come from you.2 and earlier) or no ip finger (IOS 11. instead of the usual logging access list command (such as access-list 106 deny ip any any log). Shut down unused interfaces using the shutdown command. Check them with the show ip interface brief command. [Section 4. For reasons of efficiency.6 East(config)# access-list 51 deny any log 5. Cisco IOS doesn't look at an entire packet header unless it has to. If the router has an auxiliary console port (aux port) and it is not in use.9. This may be the default on some recent versions of IOS.3 and later).the finger daemon. shut it down as shown below.Appendices • Finger .2.1] Central(config)# interface eth 0/3 Central(config-if)# shutdown Central(config-if)# exit Central(config)# line aux 0 Central(config-line)# no exec Central(config-line)# transport input none Central(config-line)# exit 4. Turn off this little-used server with the command no ip bootp server. a Smurf attack can let an attacker use your network to launch denial of service raids on other sites. If packets are rejected by an access list filter for other reasons. To prevent this.2. [Section 4. use the special port range arguments shown below.2] • NTP . [Section 4. If NTP is not being employed for time synchronization. Use the command no service finger (IOS 11.5] • BOOTP – the IP bootp server. NTP can also be disabled for only a specific interface with the ntp disable command. Don't be a Smurf buddy! While the Smurf attack doesn't usually attack the router itself. [Section 4.2] 2. Log access list port messages properly. no access-list 106 access-list 106 deny udp any range 0 65535 any range 0 65535 log Version 1.1c 275 . To prevent this from happening. the log message will often list the packet as using “port 0”. but include it in your configuration explicitly anyway. [Sections 4. use the command no ip directedbroadcast on all interfaces.3] East(config)# no access-list 51 East(config)# access-list 51 permit host 14. turn it off with no ntp server.

turn it off with the command no snmp-server.255 Controlling authentication for login to the router is an extremely important topic.6 for guidance. use the password command. Turn off SNMP trap authentication to prevent a remote SNMP system shutdown request. are permitted. it helps identify mis-configured internal hosts and 276 Version 1.2. [Section 4.Router Security Configuration Guide access-list 106 deny tcp any range 0 65535 any range 0 65535 log access-list 106 deny ip any any log The last line is necessary to ensure that rejected packets of protocols other than TCP and UDP are properly logged. To control access.1] South(config)# line vty 0 4 South(config-line)# login South(config-line)# password Soda-4-J1MMY South(config-line)# access-class 2 in South(config-line)# transport input telnet South(config-line)# exit South(config)# no access-list 92 South(config)# access-list 92 permit 14. virtual terminals (telnet) are unprotected. [Section 4. [Section 4.0. such as Telnet or SSH.0.2] Central(config)# no ip source-route 8.2 and later use the global configuration command no snmp-server enable traps. 7.2.1] South(config)# enable secret 2Many-Routes-4-U South(config)# 10. 4. Allow only internal addresses to enter the router from the internal interfaces.0 0. If only specific methods of attaching to the VTY. In IOS 11. use an access list and the access-class command. [Sections 4.1c . consult Sections 4. Password and access protect the Telnet VTYs. If SNMP is not being used on the router. enforce this using access lists. Block illegal addresses at the outgoing interfaces. [Section 4.10. Unless the network is one of those very rare setups that needs to allow source routed packets.3] South(config)# no snmp-server enable traps South(config)# no snmp-server 9. use the transport input command to enable only those methods.3] 6. the source routing facility should be disabled with the command no ip source-route.5. To set a password.1 and 4. Besides preventing an attacker from using the router to attack other sites. By default. Make sure that the router enable password is encrypted using the strong MD5-based algorithm by using the enable secret command rather than the enable password command.

This protection should be part of the overall design for traffic filtering at the router interface attached to the external.) The example below shows how to enforce these rules on router North.0.0 0. [Section 4.0.0. Incorporate this protection into the access lists applied to interfaces connected to any untrusted networks. and use it to log errors and blocked packets to an internal (trusted) syslog host.1).) • A number of remote attacks use ICMP redirects. (Note that this may block DHCP and BOOTP services.0. • Block incoming loopback packets (address 127.255 any log Version 1.6 12. but these services should not be used on external interfaces.5] Central(config)# Central(config)# Central(config)# Central(config)# logging logging logging logging buffered trap info facility local1 14.3] • Block packets that claim to have a source address of any internal (trusted) networks.0. This approach may not be feasible for very complicated networks.2.0 0.2. Block packets coming from the outside (untrusted network) that are obviously fake or are commonly used for attacks. North(config)# North(config)# North(config)# North(config)# North(config)# North(config)# no access-list 107 ! block internal addresses coming from outside access-list 107 deny ip 14. [Section 4.255 any East(config)# access-list 101 deny udp any range 1 65535 any log East(config)# access-list 101 deny tcp any range 1 65535 any log East(config)# access-list 101 deny ip any any log East(config)# interface eth 1 East(config-if)# ip access-group 101 in East(config-if)# exit East(config)# interface eth 0 East(config-if)# ip access-group 101 out East(config-if)# end 11. Make sure that the router blocks syslog traffic from untrusted networks.0.0. (A superior but more difficult approach is to permit only necessary ICMP packet types. • Block broadcast packets. • If the network does not need IP multicast then block it.255 any log ! block bogus loopback addresses access-list 107 deny ip 127.0. These packets cannot be real. 0. Turn on the router’s logging capability.2. untrusted network.255 any log access-list 107 deny ip 14.9.1c 277 .255.1 0.3] East(config)# no access-list 101 East(config)# access-list 101 permit ip 14.0. [Section 4.255. block them. This impedes some TCP sequence number guessing attacks and related attacks.Appendices networks.0.0.

255. using a rule like the one shown below (part of the configuration file for router East). Unless the router absolutely needs to autoload its startup configuration from a TFTP host. [Section 4.250 log access-list 102 permit ip any any interface Eth 0/0 ip address 14. So that the complete date and time are stamped onto entries in the routers buffered log. If proxy ARP is not needed.255.e. [Section 4. Block incoming packets that claim to have the same destination and source address (i. [Section 4.2.2] 278 Version 1. . include the command no ip http server in all router configurations.250 host 14.0 0. using the command service tcp-keepalives-in.2] Central(config)# interface eth 0/0 Central(config-if)# no proxy-arp 16.1] 15. Incorporate this protection into the access list used to restrict incoming traffic into each interface. Except on the rarely-seen Cisco 1000 series routers. [Section 4.0. Turn on TCP keepalive packets for administrative telnet sessions.6.250 255. To be safe.3] no access-list 102 access-list 102 deny ip host 14. however. use the global configuration command service timestamps as shown in the example below.0.1c .0.0.Router Security Configuration Guide North(config)# ! block multicast North(config)# access-list 107 deny ip 224.0 ip access-group 102 in 14. Disable it by applying the command no proxy-arp to each external interface.2. the HTTP server is off by default.2] 17.0. North(config)# interface eth 0/0 North(config-if)# ip access-group 107 in 13. [Section 4. disable it on all interfaces.6.6. Proxy ARP is used to set up routes on the fly for internal hosts or subnets and may reveal internal addresses. [Section 4. a ‘Land’ attack on the router itself). disable network autoloading with the command no service config.0 any log North(config)# ! block ICMP redirects North(config)# access-list 107 deny icmp any any redirect log .5] East(config)# service timestamps log date \ msec local show-timezone East(config)# 18.255 any North(config)# ! block broadcast North(config)# access-list 107 deny ip host 0.255.2.

so plan your updates carefully.1] East(config)# service password-encryption 20. Turn on password encryption. newer releases will include fixes for security bugs. Update your IOS image to the latest General Deployment (GD) release. It is not necessary to install each and every new IOS release. In general. Version 1. [Section 4. and defending against common attacks.Appendices 19. so that regular passwords are stored and displayed in scrambled form. and will provide new security features. [Section 4. Installing an update normally imposes some downtime.1c 279 . This provides some security against casual ‘over-the-shoulder’ attacks. but it is a good idea to keep your router up to date. see Section 6.5] For more information about testing router security.

gov/ia. it is best to create a security checklist for LAN switches. hubs. switches and hubs cannot generally enforce security policies on network traffic.2. The discussion below focuses mainly on media-level network components: switches.nsa. Security Principles and Goals The general security goals for a switch or smart hub are similar to those for a router. routers were the only LAN components with sufficient flexibility to need security configuration. and bridges. an Ethernet MAC address). it is important to configure them securely. managed hubs. The security goals for a switch or hub are listed below. If possible. More information is available in [4]. but simpler because such a network component does not act as a boundary device between different networks. 8. and remote network connections. access servers. Such devices almost always support remote administration and management. and other network hardware components that are not IP routers. Since the mid-1990s. It can be downloaded from http://www. and other LAN components have acquired substantial capabilities. Prior to the 1990s. Application to Ethernet Switches and Related Non-Router Network Hardware This appendix identifies specific topical areas and recommendations from the main body of this guide that apply to Ethernet switches. and preventing their use by unauthorized individuals and attackers.2.g. The NSA “Cisco IOS Switch Security Configuration Guide” [2] provides a great deal more information on the secure configuration of Cisco Ethernet switches.Router Security Configuration Guide 8. local connections. many of them are as flexible and configurable as a router. Follow the general form of the security checklist given at the end of Section 3. 280 Version 1. managed hubs.1. access servers. The focus for security for these devices is protecting their own configuration. • preventing unauthorized examination of device state and configuration • preventing unauthorized changes to the device state and configuration • preventing use of the device for attacking the local network • preventing unauthorized remote management/monitoring of the device To achieve these goals. These devices are characterized by participation in the network itself by forwarding and switching traffic based on a media layer address (e. switches. and are therefore subject to compromise over the network. the device must be configured to strictly limit all forms of access: physical. Because they cannot perform network layer or transport layer traffic filtering. Because they are vital to network operations and because they can be used as a staging area for additional attacks.1c .

The syntax shown in Section 4 will work for IOS-based switches. Especially important for 2900 switches is restricting access to the HTTP server. .5(1) . System Bootstrap Version 5. IOS-based Catalyst 2900 sw20c# show version Cisco Internetwork Operating System Software IOS (tm) C2900XL Software (C2900XL-H-M).2(8)SA. sw20c uptime is 6 days.2. and configuration loading and maintenance. .1(2) .2. and thus does not need to be configured. . Table 8-1: Router Security Guidance Sections Applicable to IOS-based Switches Section Topic Access security Application to Switches All of this section applies to switches: setting up users and passwords. Before attempting to apply the detailed instructions from Section 4.5(1) NmpSW: 4. and even some smart Ethernet hubs. Catalyst 5000 series). 4. login to the switch and use the show version command to display the operating system name. Much of the security guidance given in Section 4 that can be applied to IOS-based Cisco switches.2 Network service security Version 1.1 4. While the command syntax and command interface structure differ between Cisco IOS-based and other equipment. RELEASE SOFTWARE (fc1) . . remote access restrictions. 2900 series) and those that do not use IOS (e. any network service that is related to routing usually is not supported on a switch.1c 281 . In addition.Appendices 8.g. 51 minutes Cat5k# The table below describes how to apply the guidance in each part of Section 4 to IOS-based LAN switches. the same general principles apply to all of them. sw20c# Non-IOS Catalyst 5500 Cat5k# show version WS-C5505 Software. 3 hours. the operating system name and version are underlined in the examples below. If you do not have the switch documentation handy. check whether the particular switch is running IOS or some other operating system. Application to Cisco IOS-based LAN Equipment Cisco makes several kinds of network switches. but they can be divided into two broad groups: those that use Cisco IOS or a derivative (e. Version 11.g. Uptime is 45 days. Most of the recommendations in this section apply to switches. 9 minutes . but will not generally work on other devices. all ports should be configured to block traffic to unknown addresses using the port block interface configuration command. Version McpSW: 4. 3 hours.

VLAN configuration. but not for filtering traffic passing through the switch. June 2004.6 Access control with AAA Note that Cisco switch-resident routing hardware (e. although some Cisco switches can act as routers. after careful consideration of its role in the network security policy.2. [Note: some Catalyst 5000 and higher series switches are equipped with a ‘Route Switch Module’. on a switch. All of this section is applicable to IOS-based switches. GIAC Paper. if they support AAA (IOS 11. NSA. A. Basically. References [1] This highly detailed guide covers general switch security. Duesterhaus. Catalyst 5000 series Route Switch Modules) can and should be configured using the guidance in Section 4. too.pdf An examination of the security (and lack of it) provided by separating traffic with Ethernet switches. and must have their time set manually. includes several good references.. but the discussion is still valuable.1c .Router Security Configuration Guide Section Topic Access lists Application to Switches IOS-based switches support IP access lists. 282 Version 1.2 and later).3. December 2000. [2] Borza. access lists are used for limiting access to services on the switch itself. these services should be disabled if not in use. It should be configured using Section 4 like any other router. et al. Consult [2] for more detailed guidance and specific examples. This module is essentially a 4700-series IOS router attached to the switch. or access to them should be restricted. and more. port management.] Almost all of this section applies to IOS-based switches. available at: http://www. but do not use them for as many different purposes as a router does.g. “Network Insecurity with Switches”.3 4. Many of the issues discussed in the paper have since been fixed. This section is not usually applicable to switches. 8. available under: http://www.5 Audit and Management 4. Most of the security testing guidance given in Section 6 also applies to LAN switches.nsa. All switches support RMON and SNMP. “Cisco IOS Switch Configuration Guide”.4 Routing protocols 4.synfin. some switch IOS versions do not support NTP.

3T” is IOS version 12. Some of the suffixes that you might see on special-purpose releases include “XA”. Overview of Cisco IOS Versions and Releases Cisco provides a very large number of software releases for their routers and other products.1c 283 . You might also see maintenance revision numbers in parentheses. The current format for a Cisco IOS release name is shown below. VV.3 Release = 12. especially during the flurry of bug fix releases that tends to follow a major change. for example.Appendices 8. the release “12.3. An early release will usually include an identifier. it is usually best to avoid early release software. This appendix provides an overview of the major release levels.0.2(9)XA. identifier T.0 Revision = 3 Release = 11. the format has changed several times since IOS was first introduced in the mid-1990s. such as 12.1. There is a complex naming scheme for early releases that is beyond the scope of this guide. It is intended to help with upgrade strategies and version selection.1.N.3 Revision = 5 Identifier = T 11. to understand what Cisco provides. revision 3. 11. For operational purposes.3.M RR Release identifier Maintenance revision number Minor release number IOS Major release number: 10. use this appendix and the materials listed in the references. “F”. Version 1. 12 Examples: 12. A regular release will almost always have a simple number with no release identifier. release number and release identifiers tell what features could be available. 11.8. and the revision number tells how many times the release has undergone fixes to correct problems. Often it will not be practical to install all the updates that Cisco makes available. usually for ED releases. Devise a consistent upgrade strategy that matches the needs of your network. For example. operational routers should kept up to date with the newest stable release that provides all the needed features.0. consult [1] for complete details. “HA”.5T Figure 8-1 – Cisco IOS Release Naming In general. and then follow it. The ‘T’ identifier designates an early release of new technology features. critical feature. Release Levels and Names Cisco follows strict naming schemes for IOS releases. Cisco releases may be broadly divided into kinds: regular shipping releases (general or limited) and early releases. unless it has some required. 8. and the release naming scheme.3. and may also include a number in parentheses. In general.1. Unfortunately.

Features. two common names are “IP PLUS” and “ENTERPRISE/APPN”. it supplies the hardware and memory requirements for that release before permitting you to download it. and allows you to select items of interest (hardware mode.4 was LD. even as the release status moves from ED to GD. The most stable type of release.1c .1 was ED.0: for the 3640 router. ED LD GD DF Deferred Release – a release that was built and named. The table below describes the types. including Cisco IOS releases.Router Security Configuration Guide Every Cisco IOS release has a release type. so it is generally a good idea to use the simplest feature set that satisfies all of the network’s operational and security needs. protocols. release number) in any order. a GD has usually been subject to several rounds of bug fixes since first shipping. Limited Deployment – this is the status of a release when it is first shipped to customers (FCS). Releases. IOS versions with more features require more memory. to give the greatest latitude for future expansion. All feature sets support basic IP routing and filtering. Most Cisco routers sold come with a GD release pre-installed.0. General Deployment – a stable shipping release suitable for general use. When you use the IOS Upgrade Planner to select a particular IOS software release. but later retracted. As an example. but have not undergone the extensive customer shakedown and bug fixes of a GD release.0. 12. it supports choosing versions in a very flexible way. The Cisco web site provides a “Software Center” where authorized customers can download software products. The part of the software center that contains the IOS releases is called the “Cisco IOS Upgrade Planner. Type Description Early Deployment – a pre-shipping release that supports new features. LD releases are usually stable. The revision numbers for a given release run sequentially.0. or telephony. Some commercial organizations customarily purchase routers with the maximum memory capacity pre-installed. look at IOS 12. DF releases are not available to customers. but some also support firewall or IPSec functions (see Section 5) or mainframe protocols.” Registered Cisco customers with software maintenance contracts may download IOS releases via the Upgrade Planner. Be very careful to check these requirements against the router on 284 Version 1. 12. It presents the different available releases in a friendly tabular arrangement. and 12. feature set. Releases at this level are sometimes pre-installed on routers sold by Cisco.8 was GD. The feature sets have names that are roughly evocative of what the features are. or hardware. and the Cisco IOS Upgrade Planner Every Cisco IOS release is built with a variety of feature sets. Remarks This could be considered the ‘beta’ release for an IOS version.

1 The 11. Ensure that amounts of installed memory meet or exceed the requirements before attempting to load the IOS release.3. 3660). The lists below describe some of the major features introduced into IOS in each of these releases. it did not offer extensive security features.5) • The IOS web server and web browser management interface [IOS 11. While exceedingly stable and robust. and engineering development for it was dropped in 1999. Using this tool [3].1 was first deployed in 1996. IOS 11. are now mostly unsupported by Cisco. and 12.2 The 11.1 is available as a GD release for all older Cisco routers. Major Releases and their Features There are at least five major releases of Cisco IOS software currently in use in operational environments: 11.1c 285 . but is not available for some of the popular newer models (e.Glossary which you hope to run the software. see Section 4.1.3. freely available on their web site.2 release was the first IOS version to fully implement Cisco’s modular architecture for router software. 8. All earlier Cisco IOS releases.g. Some of the important features • RIPv2 (see Section 4.5) • Various OSPF and BGP4 enhancements Version 1.1 release was the last IOS release to use the old ‘classic’ or monolithic architecture.2.0. 11. 7500.6) • RMON support (see Section 4.2.3) • Network address translation (NAT) • Support for RSVP and IP Quality-of-Service (see Section 7. a few of them are listed below.1. 12. • Named access control lists (See Section 4. with emphasis on security-relevant features. and earlier.3.2. A great many new features were added to IOS over the lifetime of 11.1(5) and later] • RADIUS support (as part of AAA. although some of them are still available for download. 12. you can check what IOS releases are supported on your router model.5) • Lock-and-Key dynamic access lists IOS 11. Cisco also offers a hardware/software compatibility matrix checker. 1605. IOS 11. 12.2. IOS 11. 11.

3 11.3 was used to introduce a large number of new features into IOS.2F only) • Early (pre-IPSec) VPN support • Early versions of the IOS Firewall feature set and CBAC (see Section 5.2 is available as a GD release for many older Cisco router models.3.0 are listed below.3T) • Cisco Encryption Technology (CET) VPNs • Enhancements to AAA (See Section 4.1c .4) IOS 11.3 is available for almost all Cisco router models.3T) • Reflexive access lists • TCP Intercept (full availability) • Initial support for VLAN routing • Enhanced IOS filesystem and initial support for FTP • HTTP authentication for the IOS web server IOS 11.7) • Full IOS firewall feature set and CBAC (11. but it was never officially shipped as a GD release. 12.0 was designed to be the basis for future router software releases.0 and 12.0 The 12. but only at the ED and LD release levels. IOS 11.1 through 11. IOS 12.Router Security Configuration Guide • Initial support for TCP Intercept (IOS 11. Some of the security-relevant features introduced or consolidated in 12. • Full support for the Firewall feature set and CBAC • Initial version of IOS Intrusion Detection (IDS) • Full support for IPSec • Commented IP access list entries • Full support for the Layer 2 Tunneling Protocol (L2TP) 286 Version 1.3 are listed below. • Initial implementations of IPSec (11.0T releases brought together a wide variety of features that had previously been available only in selected LD and ED releases of IOS 11. and to help eliminate the confusion of specialized releases that plagued 11. Some of the features introduced in 11.

3 The 12. • Secure Shell Version 1. simplifies basic secure configuration • New features added for password and authentication controls Version 1.5 (SSHv1) supported in nearly all IOS editions • Cisco Auto Secure feature added.1 are listed below. and many other Cisco hardware products. it was the latest release available at GD status for some devices.2 The 12. IOS 12.2 are listed below.0 is available in both LD and GD forms for almost all supported Cisco router platforms.1c 287 .1. • Enhanced IPSec certificate management and AAA integration • AAA enhancements: server groups. Some of the security updates in 12. As of summer 2005. • Improved support for IP Quality-of-Service and RSVP • Multi-Protocol Label Switching (MPLS) support • Enhancements to SSH support • Enhancements to IPSec and IKE • Turbo Access Lists (some router models) • Better application of service password-encryption IOS 12. more accounting features • Unicast reverse path forwarding security enhancements • Initial broad support for Secure Shell (SSH Version 1) server IOS 12.3 are listed below.Glossary • SNMP version 3 (See Section 4. IOS 12. As of the summer of 2003.2 reached GD status in 2002. A few of the many enhancements in 12.1 The 12.0. Some of the security features that appeared in 12.6) • Time-based access lists • General availability of IP unicast reverse-path verification [Section 4. IOS 12.3 had not yet reached GD status.4] IOS 12.3 release added several substantial new security features.1 release is an incremental step forward from 12. as well as enhancements to some core security features. and also expanded the availability of some features.2 release adds some new features to 12.

. Cisco Inside IOS Software Architecture.. Cisco IOS Releases: The Complete Reference. focused on packet handling and release adds additional security features and many new protocols and functions to IOS. Cisco White Papers. CCIE Professional Development Series.jsp. References [1] Coulibaly.0. Cisco Cisco Systems.x/Support/HWSWmatrix/hwswmatrix. Murphy. 2000. 288 Version 1. available at: http://www..3. However. A very detailed examination of IOS.4 The 12. Cisco Systems. R.) [4] Bollapragada. [2] “Cisco IOS Reference Guide”. client and server. • Secure Shell Version 2 (SSHv2) supported. V. decribes features in IOS 12. available at: http://www..2.3. M. 2000.1c .cgi This interactive web page allows you to find IOS releases compatible with particular router models. (Note: may be available to registered users only.Router Security Configuration Guide • SSL support added for the IOS web server (some IOS editions) • Improvements in PKI and firewall support • Introduced Control Plane Policing (some IOS editions) IOS 12. and includes a map of releases up through 12.M. A few of the new security features are listed below. C. 2004. This highly specialized book covers the Cisco IOS release system and release history in painstaking detail.3.html This detailed web page explains the IOS release naming it only covers up through IOS 12. [3] “Hardware/Software Compatibility Matrix”.com/ITDIT/CFN The service is available to registered customers at http://tools. • Improvements to protocol inspection for the IOS firewall • Control plane traffic policing (some IOS editions) • Many improvements to PKI and certificate support • Added login password control features: retry part of the Cisco Software Advisor. 8. White. 2001. retry lockout Finding Features in IOS Releases Cisco’s web site offers a useful service called the ‘Feature Navigator’ that supports searching for features by name or release number.

ARP is standardized in RFC 826. ACL Access List AH ARP ATM BGP CAR CBAC CDP CEF CIDR Version 1. CIDR is standardized by RFC 1518. Cisco IOS depends heavily on access lists for traffic filtering. Cisco Express Forwarding – a proprietary packet transfer technology used inside most Cisco router models. LAN. CIDR replaced the old class-based IP addressing scheme. used for network backbones. and more. Classless Inter-Domain Routing .Glossary 8. IKE. ESP) Address Resolution Protocol – link-layer protocol used for mapping from IP addresses to MAC addresses. IPSec configuration. (see IPSec. and telecommunications facilities. Committed Access Rate – a traffic bandwidth control facility usable for simple quality-of-service and traffic shaping tasks. and Accounting – The advanced user access control and auditing facility in Cisco IOS 11 and 12. (See also MAC Address. Authentication Header – a part of IPSec. usually based on addresses and other information from the packet headers.4. (See also LANE) Border Gateway Protocol – an advanced exterior gateway routing protocol mostly used on backbone routers. Glossary of Router Security-related Terms AAA Authentication. the packet format and protocol for IP integrity assurance services. LANs. Not commonly used today.1c 289 . Proxy-ARP) Asynchronous Transfer Mode – virtual-circuit oriented link layer protocol. permit. Content-Based Access Control – packet inspection system used for application firewall functionality in Cisco routers. or restrict network traffic. Authorization. (See also RADIUS and TACACS+) Access Control List .See Access List A set of rules that identify. access to router services.the present standard for network address allocation and network route aggregation on the Internet. BGP version 4 is standardized in RFC 1771. Cisco Discovery Protocol – a proprietary link layer protocol that Cisco routers use to identify each other on a network.

derived from the command used to request privileged EXEC mode: enable. Domain Name System – hierarchical naming scheme used for host and network names on most IP networks.Router Security Configuration Guide CPP DHCP Control Plane Policing – a security mechanism that applies rate limiting to traffic into and out of the router’s central processor. the packet format and protocol for IP confidentiality services (see also IPSec. Dynamic Host Configuration Protocol – UDP-based protocol for assigning host network attributes. on the fly. DHCP is standardized in RFC 2131. Typically. Internet Control Message Protocol – a support protocol used along with IP for control and status message. or that crash it. IKE. ICMP is a network layer protocol that provides error messages and management capabilities in IP networks. (see also OSPF). like IP addresses and gateways. A slang expression for a privileged EXEC session on a Cisco IOS router. DNS is also the name for the protocol used to transmit and relay name information. Distributed Denial of Service – This abbreviation is used for DoS attacks that use multiple (usually hundreds or more) coordinated network data sources to attack a single victim. Generalized TTL Security Mechanism – a simple spoof rejection mechanism that uses the TTL field of the IP header to detect illicit packets. FTP is standardized in RFC 959. Extended Interior Gateway Routing Protocol – A Cisco proprietary routing protocol that includes peer authentication features. The main EGP in use today is BGP version 4. ICMP is standardized in RFC 792. AH) File Transfer Protocol – widely-used TCP-based file transfer and file management protocol. DNS DoS DDoS EGP EIGRP Enable mode ESP FTP GTSM ICMP 290 Version 1. FTP control messages are passed on TCP port 21. Exterior Gateway Protocol – routing protocol designed for managing route updates between different autonomous systems. Encapsulated Security Payload – a part of IPSec. GTSM is standardized in RFC 3682. including the Internet. Denial of Service – this abbreviation is often used for network attacks that prevent a network component from providing its operational functions. DNS is standardized in RFCs 1034 and 1035.1c .

Internet Protocol version 4 – The network-layer protocol on which the Internet is built. Interior Gateway Protocol – a routing protocol used among the routers in an autonomous system.] Internet Protocol Security – a set of standards that define confidentiality and integrity protection for IP traffic. Kerberos is standardized in RFC 1510 (see also RADIUS and TACACS+). IGP IKE IOS IP IPv6 IPSec IS-IS ISAKMP Kerberos LAN LANE Version 1. [Note: security recommendations for IPv6 will be provided in a separate supplement to this guide.] Internet Protocol version 6 – The new network-layer protocol for the future of the Internet. There are two extant versions of IP: IPv4 and IPv6. Kerberos was developed by the Massachusetts Institute of Technology as a network authentication system.1c 291 . and IS-IS. RIP. IS-IS is standardized in ISO-10589 and RFC 1195. Currently popular IGPs include OSPF. and it provides strong authentication for client/server applications by using secretkey cryptography. Local Area Network – general term for a single-segment or switched network of limited physical/organizational extent. Internet Security Association Key Management Protocol – one of the precursors of IKE (see also IKE. [Note: this guide covers security only for IPv4. I is standardized in RFC 2460. EIGRP. Internet Operating System – Cisco’s name for the modular software system that runs on their routers and many other network devices. the list of current standards (STD 1) is RFC 3700. IKE is standardized in RFC 2409. IPSec is standardized by a set of RFCs including RFC 2401. LAN Emulation – A standard mechanism for routing IP packets over ATM. IPv4 is standardized in RFCs 791 and 1883. IETF standards are published by RFC number.Glossary IETF Internet Engineering Task Force – the technical and consultative body that defines standards for the Internet. Intermediate System to Intermediate System – an OSI standard interior gateway protocol based on a link state model. IPSec). Internet Key Exchange – the standard security negotiation and key management protocol used with IPSec.

L2TP is standardized in RFC 2661.255. EIGRP) Public Key Infrastructure – mechanisms and components for management of keys. Network News Transfer Protocol – a TCP-based application protocol that usually runs on port 119. An Ethernet MAC address is 48 bits long. port 123. but usually uses UDP.0. and enrollment.255. Multi-Protocol Over ATM – A proposed standard mechanism for hosting network protocols (such as IP) over ATM. can use UDP or TCP. Non-volatile RAM – device memory that can hold data even when unpowered. IGP. In IPv4. especially Ethernet interfaces. Network Time Protocol – the standard network time synchronization protocol.255 are usually multicast group addresses. standardized in RFC 3031. OSPF is standardized in RFC 2328. Cisco routers use NVRAM to hold their startup configuration.Router Security Configuration Guide L2TP Layer 2 Tunnel Protocol – A standard protocol for forwarding lowlevel protocols over IP networks. Open Shortest Path First – an IP routing protocol that uses a linkstate distance metric. Management Information Base – the hierarchical data organization used by SNMP.0 to 225. MAC Address MD5 MIB MPLS MPOA Multicast NNTP NTP NVRAM OSPF PKI Proxy 292 Version 1. certificates. Message Digest algorithm 5 – a widely-used cryptographic checksum algorithm. (See also RIP.1c . addresses from 224. An application that acts as an intermediary in the network exchanges between applications or services. Media Access Control address – the link layer address of a network interface. Proxy servers are often employed to moderate exchanges through a firewall. standardized in RFC 1321. in which packets can be broadcast to particular recipients based on address. An operational feature of IP. NTP is standardized in RFC 1305.0. (See also SNMP) Multi-Protocol Label Switching – a standard mechanism for transferring packets over backbone networks by tagging them with labels.

and/or 1812. BGP) Resource reSerVation Protocol –standard protocol for requesting quality-of-service guarantees in IP networks. The Remote Authentication Dial-In User Service (RADIUS) is specified by the IETF RFC 2058. SNMP uses UDP ports 161 and 162. TACACS+ is defined by Cisco. proposed standard. RADIUS normally uses UDP ports 1645. Secure Shell – a remote access protocol that provides end-to-end confidentiality and integrity services. RIP is standardized by RFCs 1088. SNMP is standardized in RFC 1157 and other RFCs. Terminal Access Controller Access Control System Plus – a security protocol to provide centralized authentication. Syslog usually uses UDP port 514. Secure Copy Protocol – a file transfer protocol based on SSH. Request For Comments – a document describing an Internet standard. Simple Network Management Protocol – datagram protocol used for monitoring and configuring network devices. OSPF. RADIUS support centralized authentication and accounting. or information related to or supports a standard. Proxy ARP is rarely used. and 1723. 1388. SMTP is standardized in RFC 821. (See also RIP. and accounting of users accessing a router or access server. 1646. RSVP is standardized in RFC 2205. authorization.1c 293 . (See also RMON). RADIUS RFC RIP RMON Routing RSVP SCP SMTP SNMP SSH Syslog TACACS+ Version 1.Glossary Proxy-ARP A facility offered by some routers where a router responds to ARP queries from a connected LAN on behalf of hosts on other LANs. (See also OSPF) Remote MONitoring – facilities for remote performance and traffic monitoring of network devices. A simple UDP protocol used for logging by Unix systems and Cisco routers. based on SNMP. (see SSH) Simple Mail Transfer Protocol – a TCP-based protocol for sending and relaying e-mail messages. Direction and management of paths through a multi-segment network. Use SSH instead of Telnet whenever possible. (See IETF) Router Information Protocol – a simple inter-gateway routing protocol that uses hop count as its distance metric.

Router Security Configuration Guide


Transmission Control Protocol – connection-oriented data protocol used with IP. TCP supports a large number of application layer network services, including Telnet, web, FTP, and e-mail. A simple TCP-based protocol for remote login, usually on port 23. Also used to refer to client applications that support the protocol. Trivial File Transfer Protocol – simple UDP file transfer protocol, with no authetication features. TFTP normally uses UDP port 69; it is standardized in RFC 1350. User Datagram Protocol – message-oriented data protocol used with IP. UDP is the basis for many core network services, including DNS, RIP, and NTP. UDP is standardized in RFC 768. Virtual LAN – a link layer communication domain that spans several link layer switches; commonly used with Ethernet switches. Virtual Private Dialup Network – an application of VPN technology to secure remote-dialup connections, giving a remote user secure connectivity to their ‘home base’ network. (see also VPN) Virtual Private Network – a closed network of computers or LANs, using the public network as the transport. Usually, traffic between members of the VPN is protected by IPSec during transit over the public network. Virtual TeletYpe – an interface on a host or router that provides the interactive services of a terminal. Cisco routers use VTY lines to host Telnet sessions (see Telnet).

Telnet TFTP





Cisco offers a large glossary of networking terms and acronyms at their web site: Information about a wide variety of protocols may be found at Internet RFCs are available from


Version 1.1c

Additional Resources

9. Additional Resources
The references below can be useful in designing secure network configurations, and in understanding and maintaining router security.

9.1. Bibliography
The list below consists of books that are useful for router configuration and security, collected from the reference lists throughout this guide. Akin, T. Hardening Cisco Routers, O’Reilly Associates, 2002. A very good prescriptive guide to securing Cisco IOS routers. Albritton, J. Cisco IOS Essentials, McGraw-Hill, 1999. An excellent introduction to basic usage and configuration of IOS routers. Ballew, S.M., Managing IP Networks with Cisco Routers, O’Reilly Associates, 1997. A practical introduction to the concepts and practices for using Cisco routers, with lots of pragmatic examples. Baker, F. ed. “Requirements for IP Version 4 Routers”, RFC 1812, June 1995. A comprehensive introduction to the facilities that an IP router must provide to support the Internet. Black, U. IP Routing Protocols, Prentice Hall, 2000. A very good survey of routing protocols and the technologies behind them, with some discussion of applications. Buckley, A. ed. Cisco IOS 12.0 Configuration Fundamentals, Cisco Press, 1999. This is the reference manual and guide for basic configuration tasks in IOS 12.0. Sections particularly relevant to Router Access Security include: IOS User Interfaces and File Management. Chapman, D.B., Cooper, S., and Zwicky, E.D., Building Internet Firewalls, 2nd Edition, O’Reilly & Associates, 2000. A seminal overview of network boundary security concerns and techniques. This revised edition includes extensive updates for newer technologies. Chappell, Laura, Editor, Advanced Cisco Router Configuration, Cisco Press, 1999. Great reference book for a variety of Cisco configuration topics, including routing and routing protocols.

Version 1.1c


Router Security Configuration Guide

Cisco IOS 12.0 Configuration Fundamentals, Cisco Press, 1999. The configuration fundamentals guide and reference in book form; handy to have, but the documentation CD is usually easier to use. Cisco IOS Release 12.0 Security Configuration Guide, Cisco Press, 1999. This is the reference manual and guide for major security features in IOS 12.0, along with many examples. Desmeules, R., Cisco Self-Study: Implementing Cisco IPv6 Networks, Cisco Press, 2003. A sound introduction to IPv6 concepts and protocols, with directions for configuring Cisco IOS routers. Doraswamy, N. and Harkins, D. IPSec: The New Security Standard for the Internet, Intranets, and Virtual Private Networks, Prentice-Hall, 1999. Contains a good overview and substantial technical detail about IPSec and related topics. Doyle. J., Routing TCP/IP - Volume 1, Cisco Press, 1998. Offers a very detailed examination of routing protocols and design of efficient networks. Held, G., and Hundley, K., Cisco Access List Field Guide, McGraw-Hill, 1999. This book offers detailed information and examples on access list syntax and usage. Held, G. and Hundley, K., Cisco Security Architectures, McGraw-Hill, 1999. This book includes excellent general advice about router and router-related network security, in addition to its Cisco-specific material. Huitema, C. Routing in the Internet, 2nd Edition, Addison-Wesley, 1999. A deep and detailed textbook about IP routing technologies, protocols, and how routing works in the Internet Moy, J.T. OSPF – Anatomy of an Internet Routing Protocol, Addison-Wesley, 1998. Detailed analysis of OSPF, with lots of practical advice, too. Includes a good section on troubleshooting. Parkhurst, W.R. Cisco Router OSPF - Design and Implementation Guide, McGraw-Hill, 1998. Comprehensive and practical guide to OSPF use. Includes discussion of design issues, security, implementation, and deployment.


Version 1.1c

Additional Resources

Rybaczyk, P., Cisco Router Troubleshooting Handbook, M&T Books, 2000 A very practical book, oriented toward finding and correcting problems with router connectivity and routing protocols. Sedayao, J., Cisco IOS Access Lists, O’Reilly Associates, 2001. Provides detailed information on constructing and using access lists. Shamin, F. Troubleshooting IP Routing Protocols, Cisco Press, 2002. A comprehensive hands-on guide for resolving IP routing problems Stevens, W.R., TCP/IP Illustrated, Volume 1, Addison-Wesley, 1994. The most comprehensive and readable guide to the TCP/IP protocol suite; great technical background for any network analyst. Thomas, T.M. OSPF Network Design Solutions, Cisco Press, 1998. This book starts with a good overview of IP routing and related technologies, then goes on to explain how to configure Cisco routers for OSPF in a wide variety of situations.

Version 1.1c


Router Security Configuration Guide

9.2. Web Site References
The list below consists of pointers to web sites that provide useful information about routers, network security, and vulnerabilities. CERT

The Carnegie-Mellon University Computer Emergency Response Team (CERT) maintains a web site about network vulnerabilities. Many of the incident reports, advisories, and tips are relevant to router security. Cisco Documentation

This is the root of the Cisco documentation tree. From this page, you can find IOS software documentation, tutorials, case studies, and more. Cisco Press

At the web site of Cisco’s publishing arm, you can order a wide variety of books about Cisco routers and related networking technologies. Cisco Technology Support: Security and VPN tsd_technology_support_category_home.html

This page is the root of Cisco’s security technology support area. From here, you can find papers about a wide variety of security products and protocols. IETF

The IETF is the standards body that defines and maintains the protocol standards for the Internet. Use these sites to look up protocol standards and track emerging technologies that are becoming standards. Microsoft

Microsoft’s site offers extensive information about networking their products, and about product vulnerabilities. This information can often be helpful in configuring routers that protect Microsoft-based networks.


Version 1.1c

Additional Resources

Packet Storm

This site is a good resource for network security news, vulnerability announcements, and especially testing and attack tools.

This commercial web site offers descriptions and links to information about a very wide range of protocols and telecommunication data formats, as well as a pretty good glossary. Security Focus

Security Focus is a good site for security news and vulnerabilities. Although it doesn’t usually have much information about routers, it sometimes gives advice on how to forestall certain attacks by using your routers.

Version 1.1c


sourceforge. earlier versions should not be upgraded. NET-SNMP http://net-snmp. a network traffic capture and analysis tool. flexible. While Nessus is primarily oriented toward scanning host minicom is fast. Minicom http://www.ethereal. efficient. originally created and distributed by the University of California at Davis.9. and can serve as an effective Cisco router console application on The FreeRADIUS server is a highly configurable open-source RADIUS server implementation. While it has no fancy GUI.1. Kiwi Syslog Ethereal is an effective “sniffer”.net/ NET-SNMP is a free software toolkit for SNMP.Router Security Configuration Guide 9. Tools like Ethereal are valuable for diagnosing and testing router and network security. effective terminal emulation tool for Linux and The Nessus security scanner is a handy tool for getting a quick idea of the security vulnerabilities present on a network.kiwisyslog.freeradius. 300 Version 1. Tool References The list below describes some available commercial and non-commercial tools that may be helpful in router administration and improving network security.3.1c . FreeRADIUS http://www.html Minicom is a small.pp. Nessus http://www. It was formerly called “ucd-snmp”. The current version is 0. The Kiwi Syslog is one of several freely available syslog servers for Windows operating systems.clinet. it may also be used to scan A syslog server is necessary to capture and preserve log messages from Cisco routers and many other network devices. Ethereal http://www.

SAINT http://www. and more. and is an invaluable tool for confirming filtering configurations. OpenSSH http://www. with benchmark rules specifically targeted to checking Cisco IOS and address scans in a variety of RAT is a tool for checking network device configurations. It makes an effective Cisco router console This open source tool provides a variety of services for managing operational routers: configuration difference The OpenSSH project offers a free. routers. but it can also identify security vulnerabilities of routers and the network boundary protection they provide. RANCID http://www.eeye. TeraTerm Pro http://hp.html The Security Administrator’s Tool for Analyzing Networks (SATAN) is primarily oriented toward network security assessment of traditional host 301 .org/nmap/ and networks.Additional Resources Nmap Version 1. RAT http://www.insecure. configuration change tracking. SATAN http://www. mass application of commands.openssh. usable implementation of the SSH security protocol for a wide variety of The Security Administrator’s Integrated Network Tool (SAINT) is an advanced derivative of SATAN. The included rule sets may be used.cisecurity. It can perform TCP. UDP. It can provide valuable security scanning services for hosts.html TeraTerm is a freely available terminal emulator and telnet application for Windows operating systems. Version 2.html This is the most widely used port-scanning tool for Linux and Unix systems. or extended with rules that enforce your local security RAT is provided by the Center for Internet Security.1 was the latest available at the time this guide edition was published. A version is also available for Windows NT/2000 systems.

77 ntp. 228 debug. 187. 84 IPSec and. 58 302 Version 1. 275 SYN flood. 119 Accounting. 239 and access lists. 166 show rmon. See AAA timeout for SSH. 179 RADIUS. 98 Cryptography. See BGP route flap damping Data Link Layer. 195. 91. 252 Attacks. 68 examples. 193 access-class. 183 Kerberos. 252 Console. 97. 59 logging. 59 bgp dampening. 72 service udp-small-servers. 195 TACACS+. 75 service pad. 67 show snmp. 192 server groups. 180 aaa server group. 124 Black Hole Routing. 98 radius-server. 37 Domain name service Kerberos and. 176. 77. 83 banner. 145. 265 proxy ARP. 168 Committed Access Rate (CAR). 109 router rip. 163 shutdown. 60 username. 215 DMZ. 98 clock. 73 ip inspect. 143 trace ip. 266 Attack Tests. 59. 115 kerberos. 61 show logging. 83–93. 150. 95 rmon. 252. 173 distribute-list. 127 TTL Security. 239 viewing current. 79. 253 testing with. 25. 194 BGP. 95 extended. 193. 90 password sniffing. Null routing C CBAC. 60 reacting to. 194 terminal. 184 authentication. 15 martian. 114 router ospf. 172. 257 smurf Attack. 112 line. 99 copy. See Users Address CIDR allocation. 83 URPF and. 15 MAC address. 196 resolution. 147 config t. 98 Conventions. 41. See AAA Auxiliary port. 228 ip verify. 193 rate-limit. 22 AS numbers. 229 servers. 177. 154 show version. 60 logging. 181 login local. 10 CPU. 228 crypto maps. 207 exec-timeout. 180 authorization. 239 Abbreviations. 72 CEF. 79 ip audit. 60 access-list. 111 ip ssh. 18 Debugging. 161 snmp-server. 125 router eigrp. 183 aaa new-model. 72 control-plane. 237 testing. 62 encryption. 228 D Dampening. 50. 28. 181 aaa authorization. 175–200 accounting. 233 ARP. 198 key chain. 154 tacacs-server. 68. 60 mop enabled. 25. 10 Access Control Lists. 157 router bgp. 126 spoofing. 99 show crypto. See also AAA Accounts. 95 Configurations auto-loading. 35. for IPSec. 66. 21 Commands aaa accounting. 170 ATM. 271 SSH and. 125 transport. 184 aaa authentication. 227 Authorization. 276 and VTYs. 160 show ip. 75. 126 route flap damping. 57. 215 named. 178. 59 Application Layer. 127 ip rip. 75 service tcp-keepalives-in. 195–200 method lists. 84 standard. 60 hostname. 236 global parameters. 142. 60 CAR rules. 142–45 login authentication. 88 enable secret. 168 show line. 72 service-policy. 209 Logging and. 56.1c . 78 Secure DNS. 278 table. 160. 123 prefix filtering. See Logging Authentication for IKE/IPSec. 212 show flash. 93. 29 DDoS. 110. 206. 253 Denial of Service. 145 mirror. 127 Banners. 132 isis authentication. 72 class-map. 96 CBAC and. 206–11. 22 IP. 161 crypto. 161 ip http. 117. 211 Audit. 132 CDP. 140 Control plane policing. 236 compiled. 75 copying. 112 router isis.Router Security Configuration Guide Index A AAA. 242 ip cef. 150 passive-interface. See Routing. 105. 110 service config. 131 CIDR. 79 ip ospf. 66 service tcp-small-servers. 57 connect. 109 ip prefix-list. 67 configuring. 157 show running. 132 ip ftp. 88 policy-map. 252 Land attack. 60 B Backbone Router. 128 cdp. 208 retries. 237 ip name-server. 79. 180.

122 Version 1. 166 R RADIUS. 105 1700. 109 for IKE. 71. 40 route advertisements. 15 icon for. See IPSec Interfaces ACLs on. 194 key management. 104 authentication. 7. 63. 140 IDS Post Office. 19 Ethernet. 24 updating. 170 MPLS. 265 IS-IS. 88–93. 97. 267 N NAT. 33. 39 exterior gateway. 235 setting username. 193 for SSH. 104 authentication for. 243 M MD5. 253. 27 NTP authentication. 205–17 F Filtering. 105 Intrusion Detection (IDS). 235 TACACS+. 111. 145 Syslog logging. 27 Network Layer. 192 routing protocols. 18 address. 11 IGP. 20 MAC address. 91. 63 release numbering. 228 for TACACS+. 160. See Keys Keys. 74 L LAN. 64 SSH for. 151 viewing logs. 58 CBAC and. See AAA RAM. 123 timers. 73 over SSL. 129 passive (routing). 63 G Group SNMP. 277 interface. See CBAC message format. 194 H H. 275 viewing status. 142 IOS AAA and versions. 79. 118 Filters. 113 OSPF. 88.Index E EGP. 123 IKE. 56 RAM. 15 NVRAM. 168 SMTP. 51 1757. 156 1918. 67 dial-in. 62 encrypting. 283. 124 826. 103 EIGRP. See AAA Key chains. 142. 232 Network Layer. 126 Privileges. 85 binding. 160. 169 Q Quality of service. 76 redirects. 121 strategies. 146 Syslog host configuration.1c 303 . 88 2267.323. 103 mop. 205 for OSPF. 150 configuring. 19 compare to IPv6. 43 CET. 145 timestamps in. 161 K Kerberos. 160 FLASH. 243 Message severity levels. 7. 190 IPSec for. 254 2865. 204 for IPSec. 57 timeout. 241–46 Director. 77 IP Addresses. 284 versions. See Access Control List Firewall. 59 O OSI Model. 49 and RIP. See IOS RIP. 24. 38. 270 Legal notice. 55 Prefix lists. 103 IKE. 207 Memory. 264 I IANA. 75 Ethernet. 61 Protocols by port number. 217 testing. 285–88 viewing processes. 213 IPv6. 276 address. 232–40 FTP. 16 addresses. 57 Firewall. See Memory Remote Administration. 104 authentication for. 44. 229. 217 rules for. 139–46 destinations for. 18 Physical Security. 115 NTP authentication. 125. 15. 238 IPSec and. 108 for RADIUS. 46. 102. 5. 283 role of. 149 disabling. 106 migration to. 158. 103. 76 source routing. 112 Encryption. 124 Physical Layer. 109. 281 command modes. 141 SNMP trap logging. 279. 279 rules for. 18 IPSec. 235 disabling. 120 unused. 88. 68. 58 RFC. 11 switches. 227 loopback address. 281 virtual. 24 EXEC mode. 104 distribute lists. 154 GTSM. 77 RADIUS. IOS and. 265 directed broadcast. 17 1027. 15 Architecture. 192 3031. 256. 265 Icons. 267 3682. 76. 180 and switches. 150 P Passwords. 59 Logging. See SNMP Modes IOS modes. 51 ICMP. 207 of passwords. 234 HTTP. 268 for administration. disabling. 17 Layers. 204 interior gateway. 141 MIB. 38 dynamic. 144. 212 loopback. 58 null. 28.

72 timestamps. 86 nmap scan. 253 configuring. 231 TeraTerm. 145 using SSH on. 72 V VLANs. 76. 65 reverse telnet. See VTY Services bootp. 110 Software Updates. 70. 227 availability. 1. 60 protecting. 231 diagnosing. 251 Users. 86 nmap scan. 238 IKE port. 62 viewing logged in. 166–73 management of. 167 SSH. 5. 76 disabling. 132 Unix finger protocol. 43 neighbor authentication. 130 viewing. 39 small servers. 85 intercept. 258 SAINT. 256 SATAN. 153 RMON. 16. 90 keepalives. 27 Turbo ACLs. 251 NTP port. 59 dial-in. 230 RAT. 301 304 Version 1. See AAA TCP and CAR. 73 HTTP. 251 TFTP. 233 handshake. 72 SYN attack. 205 in access lists. 129. 152 Tools CyberCop. 148 RANCID. 163 RMON. 75 TCP small servers. 160 Time. 70. 255 functional. 69 Security for the Simple Network Management Protocol (SNMP). 158 procedure for. 6 T TACACS+. 73 NTP. 88. 204 security policy for. 59 VTY line. 59 for SSH. 231 SHA. 106 role of. 221 SNMP. 67. 74 CDP. 146–50 time zone. 254 traceroute. 130.Router Security Configuration Guide migration from. 140. 60 ACLs on. 263 VPN. 95 Switches. 171. 267 U UDP. 251 port numbers. 46. 228 RSVP. 121 dynamic. See IPSec VTY. 251 SCP. 87 log messages on. 73 syslog. 250 Router Audit Tool (RAT). 27 and CBAC. 270 Unicast RPF. 149 open sockets.1c . 68 testing. 77 source routing. 72 syslog port. 280 Switching. 103 null routing. 9. 66 logging port numbers. 64 defining local. 276 ACLs on. 148 list of. 97 and CBAC. 263 CEF.1. 161 security concerns. See RMON v3 security levels. 196 SSH clients. 123 version command. 143 number of. 169 port numbers. 151–55 Security Policy. 44. 46 security testing. 155–58 Route Table. 238 for EXEC. 103 distribute lists. 168 TCP/IP. 35. 228 W Wide-Area Network (WAN). 235. 287 clients. 274 finger. 106 URPF Verification. 252 automated. 68 nmap. 165 transcript of. 28. 190 passwords. 87 configuring. 276 viewing status. 126 static routes. 41 in access lists. 230 S Scanning. 119 prefix list filtering. 70. 48 Unreachables. 72 directed broadcast. 15 Telnet. 75 SNMP. 207. 102–30 default. 7. 258 diagnostic commands. 251 ntpq. 106 unicast RPF and. 84 logging port numbers. 300 minicom. 250 Transport Layer. 256 SSH clients. 129 passive interfaces. 130 Syslog. See Logging System and Network Attack Center (SNAC). 169 Testing. 77 pad. 103. 255 for NTP. 278 UDP small servers. 145. 37 Review of. 210. 250–57 attack. 90 RSA. 47. 24. 17 filters for. 39 small servers. 219 Kerberos and. 132 MPLS. 147 Timeouts for CBAC. 116 Checklist. 92 Routing. 152 vulnerability in. 16 Windows IPSec and.