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

See Section 2. Routers direct network data messages.1.9.Introduction 1. it is feasible to use simple broadcast or sequential mechanisms for moving data from point to point. Most large computer networks use the TCP/IP protocol suite. more complex networks.2. Version 1. Introduction 1.10. In larger. Router 1.2. and sends the packet over LAN 1 to its gateway. the security of routers and their configuration settings is vital to network operation. Router 2 sends the packet over LAN 3 to the File Server. Directing data between portions of a network is the primary purpose of a router.20. Consulting its internal route table.2.6. Therefore.12 Router 1 Wide Area Network LAN 2 File Server 14. based on internal addresses and tables of routes. Without robust routing.3 for a quick review of TCP/IP and IP addressing. An Ethernet local area network (LAN) is essentially a broadcast network. data must be directed specifically to the intended destination. illustrates the primary function of a router in a small IP network. it creates a packet with address 14. below. or packets. In practice.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).2. or known destinations that serve certain addresses. The Roles of Routers in Modern Networks On a very small computer network.0 User Host 190. LAN 1 190. Consulting its own route table.0 Router 2 LAN 3 14. the operation of any large network depends on the route tables in all of its constituent routers. Router 1 forwards the packet to Router 2. Figure 1-1.1c 7 .9. most modern networks cannot function.2.

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

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

D. The following typographic conventions are used as part of presenting the examples. this guide uses many abbreviations commonly employed in the Cisco documentation and literature. to make it narrower.D/N.2. the interface name ethernet is commonly abbreviated “eth” and the command configure terminal is commonly abbreviated “config t”.0.” • Sequences of commands to be used in a configuration are shown separately from the text. user input and comments are shown in Courier bold typeface.9. using Courier typeface. Input in the transcript is distinguished from output.2. Central> enable Password: Central# ! list interfaces in concise format Central# show ip interface brief Interface IP Address OK? Method Ethernet 0/0 14. ! set the log host IP address and buffer size logging 14. For example.) • Cisco IOS accepts the shortest unique. 10 Version 1. use the command show ip route.C. • 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. unambiguous abbreviation for any command or keyword.6 logging buffered 16000 • Transcripts of router sessions are shown separately from the text. this classless netmask notation will be used where a netmask is relevant.3.2.Router Security Configuration Guide 1.15.250 YES Manual . The exclamation point begins a comment line. usually a remark about the line that follows it.C. Central# exit • IP addresses will be shown in the text and in diagrams as A.1c .255.” Command arguments are shown in Courier italics: “syntax for a simple IP access list rule is access-list number permit host address. Elision of long output is denoted by two dots. For example. .2.250 YES NVRAM Ethernet 0/1 14. 14.B. In some cases.9. the bare address will be used. (In general. most of the sections include extensive instructions and examples.B. For commands that are typed very frequently.255. using Courier typeface. where N is the number of set bits in the IP netmask. or as A. output that would be too wide to fit on the page is shown with some white space removed.150/24 has a netmask of 255. Otherwise.9.

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

securing router network services.Router Security Configuration Guide 1. 12 Version 1. controlling traffic and filtering using a router. and describes some simple network security threats. but which may become important in the near future. with a focus on Cisco routers’ capabilities. readers may want to skip directly to the sections most immediately useful to them. • Section 5 describes advanced security services that some routers can provide. and a router security glossary. Secure Shell (SSH). • 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. this section describes some areas that a router security policy should address. • 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). • Section 2 reviews some background information about TCP/IP networking and network security. The main topics of this section are IP security (IPSec). • Section 6 presents testing and troubleshooting techniques for router security. • Section 9 provides a list of resources. • Section 3 presents a security model for routers. It is essential for good security that any router security configuration undergoes testing. using recent versions of the Cisco IOS software. References are included at the end of each section. along with a discussion of relationships between router security and overall network security. collected from all the sections of the guide. Structural Overview The various parts of this guide are designed to be fairly independent. The list below describes the major sections. • Section 4 details the methods and commands for applying security to Cisco routers.1c . and defines general goals and mechanisms for securing routers. and this section presents both vendorindependent and Cisco-specific testing techniques. configuring routing protocols security. security management for routers. including pointers to web sites and security tools. Security mechanisms must be applied in support of security policy. It is divided into six main parts: • • • • • • securing access to the router itself. and network access control for routers.4.

• • • • Section 8. For network security planners and system security designers. the detailed instructions for locking down a router are the most important part of this guide.2 – for instructions on configuring IPSec on Cisco routers Section 5.3 – for advice and guidance on setting up filtering Section 4. 5.1 – for advice for quickly securing a Cisco router Section 8.1 – for quick tips that will greatly improve router security Section 4.3 – 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.Introduction How to Use This Guide Several different roles are involved in securing a network. if necessary Section 3 – for the security principles behind the advice in Section 4 Section 4 – for detailed instructions on configuring Cisco routers Section 5.3 – for a quick guide to using SSH for Cisco administration Section 8. this guide offers detailed advice and direction. The paragraphs below offer roadmaps for using this guide for several different network security roles. • • • • • • • • • Section 2 – for a review.1 through 4. network. Read the sections listed below if your role is security planner or security designer. • • • • • Section 2 – for a review of TCP/IP. Read the sections listed below if your goal is to quickly lock down a router.1c 13 .1 – for explicit directions on router access security Section 4.1.3 – for an idea of what Cisco routers can do for network security Section 5 – for information about Cisco router VPN. the high-level view of router security is more important than the details of Cisco router commands.2 – for instructions on applying this guide to LAN switches Section 8.4 – for routing protocol security instructions (unless the routers are using static routes exclusively) Version 1. 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. firewall. and each may need some information about router security. and router operational concepts Section 3 – for general router security principles Section 4. Read the sections listed below if your role is network administrator.

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

1. In almost all cases. Review of TCP/IP Networking As mentioned in Section 1.1. router software architecture. and network security. it is feasible to use simple broadcast or sequential (token) mechanisms for moving data from point to point. Within a LAN.1c 15 .” 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.Background and Review 2. the reference list at the end of the section includes a long list of other useful sources of background information. large networks are actually composed of a collection of LANs that are interconnected or “internetworked”. An Ethernet local area network (LAN) is essentially a (logical) bus based broadcast network. Just above the MAC level is the Logical Link Control (802. might have 100 computers (and printers) attached to a switch or set of hubs. Standards that relate to LANs are primarily the IEEE 802.5 is the MAC standard for Token Ring. Token based LANs avoid collisions by only allowing one host at time to transmit (the host that currently has the token may transmit). why not humor the author and read on.x series. To give some meaning to the term “relatively. For instance. 2. To compensate for that briefness. though the physical implementation may use hubs (with a physical star topology). 802.3 is the Media Access Control (MAC) standard for CSMA/CD (the Ethernet standard).1) standard. router hardware architecture. addressing is done using IP addresses. For instance. data must be directed more carefully. In order to keep this section brief. A local area network is composed of a relatively small number of hosts connected over a relatively small physical area. broadcast LANs must deal with collisions. If you are still lost at the end of Section 2. on a small computer network. using 100BaseT Ethernet.1. but since it is relatively brief. Background and Review This section reviews some background information about TCP/IP networking. more complex computer networks. As one would expect. “Relatively small” is the important phrase here. 2. keep reading because much of this will be explained below. Between LANs using TCP/IP. it glosses over a lot of issues. Purpose of a Router In larger. either by preventing them or detecting them and taking appropriate action.1. If you are lost at this point. This is where routers come in.2) standard and above that it the High Level Interface (802. while 802. addressing is done with a MAC address. Routers take Version 1. Readers with a good grasp of network and router fundamentals may want to skip this section. then consider reading parts of some of the books and/or web pages listed at the end of the section. a typical office LAN.

4. it allows them to help protect computers and other network components. Routers also make decisions about where to send these packets. respectively.2.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). Part of the “conversion” process is to add a packet header. Most large computer networks use the TCP/IP protocol suite. the router learns about where various addresses are relative to itself and builds up route tables based on this information. The goal is almost always to get these packets to another LAN and ultimately to the correct host on that LAN. There are a number of schemes or routing protocols for routers to acquire and share route table information. 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. Building packets and unwrapping packets are additional router functions performed by the first and last routers. 16 Version 1. While a thorough treatment of the details is beyond the scope of this document. a router may be responsible for filtering traffic.1c . 2. there is a substantial discussion of routing protocols is in Section 4. In addition to directing packets.2 for a quick review of TCP/IP and IP addressing. allowing some packets to pass through and rejecting others.1. 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. Updating these routing tables and forwarding data packets between portions of a network are two of the primary tasks of a router. that a message passes through. 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. Other routers will generally only look at a packet’s header information. In some sense this is the lingua franca of the Internet. In dynamic routing. see Section 3 and Section 4. Filtering can be a very important function of routers. See Section 2. For more information about filtering. In the past these tables were built and updated by hand and this is referred to as static routing. 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. Route Tables As mentioned.

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

This involves such details as signal voltage and duration.2 802. It may also cover some low level error control. However. To fully understand routing it is useful to appreciate some of what goes on beneath it at the Data Link Layer. and some of this is discussed in the following sections. It also covers the mechanical aspects of connectors and cables.1 802.3 (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. 18 Version 1.Router Security Configuration Guide OSI model. A collection of various compatible protocol layers is referred to as a stack. or optical signaling details for fiber.1c . It is concerned with the transmission of an unstructured bit stream over a physical link. the Network Layer. 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. the Physical Layer is at a level of detail well below the concerns of this document.

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

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

For example. there are four “classes” of IP addresses. an organization that needed 1000 addresses would be assigned a netmask of 22 bits. The number of contiguous Class C addresses allocated is a function of the size of the network. For any class. and the other was increased route table sizes.1. There are also a set of IP addresses that are reserved for experimental or private networks. and any beginning with bits 1110 is class D.4. there is a good bit of information in an IP header. Any address beginning with a zero bit is a Class A address. the remaining bits are the host (or subnet) portion.Background and Review 2. (Another way to think of this is that CIDR allocates several contiguous Class C addresses to a network. 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. Classless Inter-Domain Routing (CIDR) and IP Addressing As the Internet grew over the 1980s and early 1990s. referred to as Classes A. In addition to both source and destination addresses. these addresses should not be used on the Internet or other wide-area networks (see Section 4. Thus it is quite feasible to introduce longer IP addresses. While many networks have more hosts than a single Class C address can accommodate (255 hosts).3). very few have enough to “fill” a Class B address range (65. it encountered two problems related to the expanding number of networks and hosts.535 hosts).3. under CIDR. One was address depletion. most notably the exhaustion of Class B networks. and D. In addition there a number of special addresses. and allocating a block of addresses using a netmask of the smallest size that satisfies the needs of the recipient.7. Moreover the second 4 bits specify the length of the header. Any address beginning with bits 110 is Class C. C. see Stevens’ book [10]. B. CIDR avoids this problem by eliminating the notion of a ‘class’. Classful IP Addressing Under the original IP standards.) Version 1. it is also possible to take the host portion of the address and further divide that range into two fields.1c 21 .3. The netmask simply specifies the number of bits in the assigned address that designate the network portion. which specify a subnet address and a host address respectively. The class is determined by the first one to four bits of the address. For a detailed explanation of TCP/IP packet header formats. 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. This is done by specifying a parameter called a subnet mask. 2. Any address beginning with bits 10 is a Class B address. Allocating an entire Class B network to an organization that only needed 1000 addresses would be (and was) terribly wasteful. For a fuller discussion of subnetting see Albritton’s book [1] or one of the other references listed in Section 2.3. 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.

255) 16384 addresses 16 addresses Customer Y 14. or consult RFCs 1771 and 1772. the definitions of standard routing protocols specify many of the details of: how routers keep track of available routes (route tables).0/13 (14.16-14. This reduces the number of entries required in the 22 Version 1.0 .0. They might allocate part of that large space as a block with netmask of 13 bits (524288 addresses) to a large ISP.14.7.0.Router Security Configuration Guide CIDR also permits address allocation authorities to allocate blocks of addresses smaller than a Class C network.2. For these topics. 16777216 addresses 524288 addresses Major ISP 14. The network aggregation required by CIDR required the development of a revised routing protocol for the Internet: BGP version 4 (BGP-4). Figure 2-4: Hierarchical IP Address Range Assignment under CIDR Internet Routing and Aggregation As alluded to in Section 2.0/8 (14.14.3. if an organization required only 10 addresses.255.0. The prefixes which specify networks under CIDR vary in length.0.0/18 (14.16/28 (14.2. A major allocation authority might obtain a block of addresses with a netmask of 8 bits (16777216 addresses).255.3. then they might be assigned a netmask of 28 bits.0.0. This essentially forms “supernets”. any meaningful discussion of routing protocols in general and BGP routing in particular is beyond the scope of this Background and Review section.24.0 . and how they decide where to forward any given packet. 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.0.1c .24. Regional Authority 14.0. which are composed of several related networks which are collectively advertised as only one aggregated path to that supernet.0.2.63. The addresses of customers X and Y would still be within the large block ‘owned’ by the major allocation authority. For example. Section 4.255) Customer X 14. In general.24. This is illustrated below. and smaller customer Y a block with netmask of 28 bits. The ISP might give big customer X a block with netmask of 18 bits. Another important aspect of CIDR is that it is hierarchical. adding a bit more complexity to routing.0 . how routers exchange this information. there is a detailed treatment in Huitema’s book [12].4 of this guide covers security issues for routing protocols.

Version 1. is put into common use these problems will continue. so that near and far do not necessarily relate to physical distances.Background and Review route tables of routers which are “far” from a given network. CIDR and BGP-4 have helped to mitigate the problem of route table size explosion. 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. Beyond CIDR Until IPv6. In the meantime. The aggregation strategy may be based on regions (geography) or providers (network topology).1c 23 .

similar to any other computer. First. specialized routing software enables a fuller and more robust implementation of filtering. but hardware other than Cisco would run similar software.g. because it can be designed to support more interface card slots. and that known vulnerabilities in other systems are of little help to the router attacker.4. But there are also a number of security benefits to a special purpose router. cost. NVRAM. Also. EEPROM). Filtering is useful as a “firewall” technique.4.4. access speed. the term IOS will be used in this guide to refer the router’s operating system and associated software. (For brevity.2. it has a central processor unit (CPU). a special purpose device will probably be a lower cost solution for a given level of functionality. and can also be used to partition networks and prohibit or restrict access to certain networks or services. ease of reprogramming. Conceptual Model for Typical Router Hardware A router is essentially just another computer. or CD-ROM drive. There are typically a number of types of memory in a router possibly including: RAM. 2. Basic Router Functional Architecture 2. concerns performance: a special purpose router can have much higher performance than a general purpose computer with routing functionality tacked onto it. So. routing protocols can advertising selected network routes to neighbors. CPU speed and memory size are important considerations for both performance and capabilities (e. one can potentially add more network connections to a machine designed for that purpose.Router Security Configuration Guide 2.1c . Flash 24 Version 1. and other factors. and connections to other devices. Typically. in general. Non-volatile RAM (NVRAM) is used to store the startup configuration that the IOS reads when the router boots. a specialized router operating system. although it may have other kinds of removable storage such as Flash memory cards. and what justifies the bother of learning yet another system? The answer. thus helping protect parts of your network.1. a router does not have a hard disk. and sophisticated security services usually require substantial computation). the mere fact that it is different means that an attacker has one more thing to learn.) This means that it is potentially less vulnerable. The mix of types and the amount of each type are determined on the basis of: volatility. Flash. like Cisco’s Internetwork Operating System (IOS) can be smaller. These are listed roughly in order of volatility. floppy drive. Using filtering. and ROM (PROM. Why Have a Special Purpose Router? What are some of the motivations for using a dedicated. Second. and more thoroughly tested than a general purpose OS. Thus. ROM is used to store a router’s bootstrap software. better understood. in part. consolidating network routing and related functions on a dedicated devices restricts access to and limits the exposure of those critical functions. purpose-built router rather than a general purpose machine with a “standard” operating system (OS)? What justifies this expense. various kinds of memory. some Cisco IOS features require more than the default amount of memory. Finally.

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

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

5. best effort protocol with no guarantee of delivery or confirmation of delivery. 2. fast-converging routing protocol. For more details see Section 4. “Routing and Routing Protocols”. the ping command relies on ICMP. 2. 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. best effort delivery of those packets. It is an interior gateway protocol that uses a link state routing algorithm and requires that a hierarchy of areas be designed. 2. When we speak of TCP/IP we are usually implicitly including UDP. RIP – Routing Information Protocol is a dynamic routing protocol that allows routers to share network information with each other.5.Background and Review 2. It is a distance vector protocol that Version 1.1c 27 . UDP – the User Datagram Protocol (UDP) is a connectionless. Review of Router-Relevant Protocols and Layers The following sections are not inclusive of all protocols that might be of interest but are representative. ARP – Hosts use the Address Resolution Protocol (ARP) to acquire the MAC address of other hosts. It has lower overhead than TCP.3. (For example. An area is a logical collection of routers and networks. Physical Layer 1 As previously discussed. Network Layer 3 IP – the Internet Protocol (IP) provides a specification for packet formatting and an unreliable.5.4. the physical layer is defined by IEEE standards or similar standards that define what are primarily physical and electrical characteristics. Transport Layer 4 TCP – the Transmission Control Protocol (TCP) is a connection-oriented.) OSPF – Open Shortest Path First is a relatively complex. Data Link Layer 2 The IEEE and other standards that apply at this layer have also been discussed previously.1. connectionless.5. The protocols are grouped according the OSI layer to which they correspond.2. reliable protocol.4. 2.5. Before transmitting data a connection must be established and after data transmission is complete the connection must be closed.

HTTP – the Hypertext Transfer Protocol (TCP) is used for retrieving web pages and many related tasks. NTP – the Network Time Protocol (UDP) is the Internet standard protocol for synchronizing time between network hosts and authoritative time sources. DNS – Domain Name System (both TCP and UDP) performs naming resolution service by translating host names into IP addresses and vice versa. 28 Version 1. 2.1c . Session Layer 5.Router Security Configuration Guide allows routers to only share information with their nearest neighbors. SNMP – Simple Network Management Protocol (UDP) enables a management station to trap certain information messages from network devices. SMTP – Simple Mail Transport Protocol (TCP) is the Internet standard protocol for transmitting e-mail messages. SSH – Secure Shell (TCP) provides cryptographic security for remote login sessions and other stream-oriented protocols. Telnet – (TCP) Enables terminal oriented processes to communicate. FTP – File Transfer Protocol (TCP) enables transfers of files between hosts. Presentation Layer 6.5. TFTP – the Trivial File Transfer Protocol (UDP) provides file transfers without any authentication or security. It is used as an interior gateway protocol. it is used for remote login. and Application Layer 7 These protocols are labeled (TCP) or (UDP) depending on which layer 4 protocol they are based upon.5.

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

routers. 1999. 1998. [5] Perlman.. O’Reilly Associates. 1992. Cisco Press. [2] Ballew. [8] Tannenbaum. J.7. Books [1] Albritton. with an emphasis on Cisco IOS. McGraw-Hill.Router Security Configuration Guide 2. S. If you need more introductory information than this section provides. K. 1997. 30 Version 1. (ed. G.1.. Interconnections: Bridges and Routers. L. and TCP/IP. Cisco Router Handbook. this book provides detailed information about a wide variety of topics and features. For the network administrator who already has basic familiarity with Cisco IOS. Computer Networks. A good book for learning the basics. this book provides concepts and practical advice for using a router securely. but it also provides a great deal of technical background information. an excellent source for understanding all the concepts behind networks. 1999. A “classic”. Cisco IOS Essentials. Introduction to Cisco Router Configuration. W. Prentice-Hall. L. This book is primarily oriented toward network application programmers. References 2. Cisco Security Architectures. McGraw-Hill. 1999. well written.7. with explanations of many of the concepts. 1998. Managing IP Networks with Cisco Routers. [3] Chappell. McGraw-Hill. 1999.. [4] Chappell. Prentice-Hall. 1998.) Advanced Cisco Router Configuration. [9] Stevens. good background reading. Unix Network Programming. McGraw-Hill. and Hundley. with no vendor emphasis.. Cisco Press. R. [6] Sacket. This book offers good explanations of all the underlying concepts. A practical introduction to the concepts and practices for using Cisco routers.. this book is a good source.R.M. [7] Held. G. 2nd edition. An excellent introduction to basic IOS operations. A. For administrators already comfortable with basic operation of a router. This thick book provides a lot of detail on the architecture of Cisco routers and their operational concepts.1c ..

1999.ietf. 1993. 1996. and . T. 1999. J. (ed.1c 31 . too. RFC 1180. [12] Huitema.2. Cisco [20] Socolofsky. includes coverage of all the topics discussed in this section. with a great deal of background information. Addison-Wesley. bit-by-bit analysis of the TCP/IP protocols. and how routing works in the A series of introductory-level papers by Cisco. Cisco Press. protocols. 2nd Edition.html 2. W. “Transmission Control Protocol (TCP)”. [14] “OSI: The Network Layer”.L. Cisco Product Overview.. these are the definitive sources for information about the protocols and architecture of the Internet. 2.R. TCP/IP Illustrated – Volume 1.0 Configuration Fundamentals. 1981. RFC 1392.. Version 1. [17] Postel. A deep and detailed textbook about IP routing technologies. 1999. Available at: http://www. [19] Baker.Background and Review [10] Stevens. “Internet Protocol (IP)”.. Cisco Systems. Routing in the Internet. Cisco Systems Brochure. [16] The Protocols. 1981. “Requirements for IP Version 4 Routers”. “Internet User’s Glossary”. RFC 1812.. “A TCP/IP Tutorial”. Prentice-Hall. J. RFC 768. RFC 793. This book provides a valuable reference for all the basic operation and configuration features.html [15] “TCP/IP”.3. G.7. Available at: http://www. For really deep. [11] Cisco IOS 12. and Parker T. “User Datagram Protocol (UDP)”. [18] Postel. F. technical. As the official documents of the Internet Engineering Task Force.. 1997. J. RFCs RFC stands for Request for Comments. 1997. Papers [13] “Internetworking Technology Overview”. All RFCs may be downloaded from http://www.). 1994. Cisco C. this book is the best source. RFC 791. C. Available at: http://www. [21] Malkin.

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

Physical Security There are a number of ways to provide physical security for a router. If continuous operation of the router is critical. or interference with router operation. an uninterruptible power supply (UPS) should be installed and spare Version 1. and generally the persistent state of the router. Router Security Principles and Goals Routers can play a role in securing networks. and generally the dynamic operation of the router. and managing a router securely. 3.Router Security Principles and Goals 3. Protecting the Router Itself 3. Threats to the management and control planes mostly concern unauthorized access to the router.1.1. Threats to the data plane usually concern violations of network security for the networks that the router supports. protecting a network with a 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. Conceptually.1. we must consider the possible threats to each plane.1c 33 . The data or forwarding plane handles the packets transiting the router among the networks it serves. The last part of this section gives some guidance about router security policy. configuration. route table updates. The management plane handles administration. The control plane covers monitoring. It should have controls for temperature and humidity. a router operates on three distinct domains or planes. The room that contains the router should be free of electrostatic or magnetic interference. This section describes general principles for protecting a router itself.

This is a false savings. Physical devices (e. Many of these services are unnecessary and may be used by an attacker for information gathering or for exploitation. * Some readers might balk at this recommendation. 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. Configuration Hardening A router is similar to many computers in that it has many services enabled by default. Also.1c . and to allow it to support the widest range of security services.1.2 discusses the management of updates to the operating system. and use the feature list to select the version of the operating system. Unnecessary services should be disabled in the router configuration. the router should be configured with the maximum amount of memory possible.3.1. 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. The incremental cost of extra memory is usually small compared to the total cost of a fully configured router. * Also. 34 Version 1. To help protect against some denial of service attacks. adding memory to an operational router requires taking that router out of service. 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. Decide what features the network needs. for example.. modems) used to connect to the router require storage protection. Section 3. 3. In the Internet Service Provider community.2. 3.g. Section 3.3.2 discusses management of updates to the router configuration. One should use the latest stable release of the operating system that meets the feature requirements. Operating System The operating system for the router is a crucial component.Router Security Configuration Guide components kept on hand. it is considered an industry best practice to equip every operational router with as much memory as it can hold. and Sections 4 and 8 include information on Cisco’s IOS operating system. PC cards.3. However. the router should be placed in a locked room accessible only to authorized personnel.

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

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

destination port. Packet filters are especially important for routers that act as the gateway between trusted and untrusted networks. This configuration offers two points at which policy can be enforced. in Figure 3-1. rejecting protocols and restricting ports according to the policies of the trusted network. In that role. It also offers an intermediate area. source port. 3. and protocol type.2.2. Some routers have filters that apply to network services in both inbound and outbound directions. while others have filters that apply only in one direction. For example. 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. the router can enforce security policy. destination IP address.g. some routers need two filters to handle bi-directional services. 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. (Many services are bi-directional. For example.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. often called the de-militarized zone (DMZ) between the two routers. The DMZ is often used for servers that must be accessible from the Internet or other external network. and System B sends some type of response back to System A. However. the router should enforce the constraint that packets sent Version 1. a user on System A telnets to System B. Some routers can even filter on any bit or any pattern of bits in the IP header. and then another router between the firewall and the trusted internal networks. Routers can apply filters in different ways.) Most routers can filter on one or more of the following: source IP address. Filters are also important for their ability to enforce addressing constraints. FTP file name). So.1c 37 . routers typically do not have the capability to filter on the content of services (e.

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

Version 1. 5000 (TCP & UDP) 2049 (UDP) 6000 .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.1c 39 . 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. the protocols listed in Table 3-1 should not be allowed across the router in either direction.

and the Defense Information System Agency (DISA) is attempting to manage the creation of a standard list for the entire DOD. or illegal address range. (Assuming that addresses are assigned correctly. and is almost certainly erroneous or malicious in nature. rdist. Address Filtering Router filters should also be used to protect against IP address spoofing. especially on border routers. including blocking reserved addresses. rcp. 40 Version 1. it is best to take the first approach. unroutable. The principles to apply on border routers are listed below. • Reject all traffic from the internal networks that bears a source IP address which does not belong to the internal networks. (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. For networks that are subject to such lists. Traffic bearing such spoofed addresses is often part of an attack. any other traffic is attempting to claim a bogus source address.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. 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. traffic sent from the external networks should always bear a source address from some range other than those assigned to the internal networks.1c . In most cases filtering rules should apply both ingress and egress filtering.) • Reject all traffic with a source or destination address belonging to any reserved. and should be dropped by a border router.) • Reject all traffic from the external networks that bears a source address belonging to the internal networks. and rejecting all others. Various organization in the US DOD maintain such lists. allowing only those ports and protocols mandated by the standard list.

2). For example.3. Of course. 1. Coordinate protective measures against distributed DoS attacks with network access providers and/or backbone administrators. and provider access. Access control lists are a general filtering facility available on all routers (see Section 4.2). 3. There are several ways to filter out these improperly-addressed packets. at both border and interior routers. Though DoS attacks can be viewed as tolerable annoyances. There are several mechanisms available on routers to thwart certain kinds of DoS attacks.3. First.1c 41 . 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. 2. 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. To be effective. they can have serious consequences if they occur during a time of crisis.3). At the enterprise level there are three primary strategies for combatting DoS attacks. Many of these attacks require use of invalid or spoofed source addresses. detection will depend on reviewing the router logs on a regular basis. these measures must be planned and in place before an attack occurs. Where possible. router services that support attacks or attack amplification should be disabled (see Section 4.3. on Cisco routers this facility is called Version 1. you should log occurences of bad packets. 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. There is no complete solution to the DoS problem. it is important for every network administrator to help reduce the number of DoS attack launch platforms. described in detail below. network devices. Mitigating Denial of Service Attacks Loss of service or severely degraded network performance can result from a variety of causes. as long as the resources of a network are limited and openly available they will be vulnerable to attack. and works on all routers (see Section 4. Black hole routing can also be useful. logging these violations can help identify compromised hosts that need to be removed from your network.6). keep hosts secure and eliminate compromised hosts from the network immediately.4. Denial of Service (DoS) refers to willful attempts to cause such disruptions. Do not let your network be the origin point for a DoS attack.7). These measures require some cooperative effort between those who administer hosts. Prevent malicious traffic from entering the common network from the enterprise network. There are measures that network administrators can take to protect networks from DoS attacks and lessen their effects. Some routers and firewalls offer specialized facilities to mitigate TCP SYN flood attacks.5).Router Security Principles and Goals 3. Also. Invalid or otherwise untraceable source addresses are often used to hide the actual source of an attack.2. Configure and deploy local protective measures.4.

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

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

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

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

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

and probably in the router security policy. In other regards. 3. the router enforces network policy. Security policy for a router should include rules about access to this layer.4. 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 that a router serves will have a security policy.1c 47 .4.2. Router Security Policy and Overall Network Security Policy Typically. The policy for a router must fit into the overall framework. and responsibilities. For example. and the router policy must reflect this. and work outwards to security for the static configuration. For example. The rules of conduct for administering the router should clarify the application of the network rules to the router. defining roles. 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. identifying permitted protocols and services. The router security policy might include only two: administrator and operator. In some cases. a network security policy might define three roles: administrator. 3. the dynamic configuration. permissions. The overall network security policy may include rules about this. Version 1. and administrative roles. The router policy might specify the particular rules to be enforced by the router to prevent unauthorized remote administration. for example. although it is sometimes overlooked. The high-level requirements of the network security policy must be reflected in the configuration of the router. the policy is more portable across router software versions and between different kinds of routers. and user. • Specify policy for all the zones identified in the figure above – Begin with physical security. The router security policy might grant the operator login privileges to the router so that they can access the router logs. The outer zone of the diagram represents the intra-network and inter-network traffic that the router manages.3. he can compromise the outermost layer as well. operator. The roles defined in the router security policy will usually be a subset of those in the network policy. the router policy will involve far more detail than the network policy. might be held responsible by the network security policy for periodic review of the audit logs. rules of conduct. rather than a particular command or mechanism. the network security policy might forbid administration of the router from anywhere but the local LAN. access mechanisms. and for traffic flow.Router Security Principles and Goals router. not particular commands or mechanisms – When the policy specifies the security results to be achieved. The operator.

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

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

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

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

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

Router Security Principles and Goals Version 1.1c 53 . Remote Remote Network 7.1.0/24 User Host 14.18/24 eth 0 East net access eth 1 14.0/24 eth 0/0 5.1/24 Mail Server 54 Version 1.250/16 Central eth 0/1 Figure 4-1: Example Network Architecture Figure 4-1 is simply a vehicle for presenting security guidance about routers. Implementing Security on Cisco Routers The diagram below shows a simple network configuration. LAN 2 14.0.0/24 User Host 14.6/24 FTP & Web Server 14.250/24 Telephone Network Public Web Server 14.1c .6. and 6.0/16 DNS Server eth 0/0 modem LAN 1 14.6/24 Admin Server 14.2/16 Facility Network Authentication Server 14. However.9.10. it is not a design for a secure network.6.2.64/24 Protected Enclave 14. this architecture reasonably reflects the kinds of networks found in many organizations.10.1.20 Internet 7.6/24 14.9.3/24 eth 0/1 (firewall) modem 14.Router Security Configuration Guide North (Perimeter router) eth 0/1 14. The structures and addresses illustrated here are used for all of the examples in Sections 4.1.64/24 Remote Host South User Host 14. eth 0/0 14.2.

1c 55 . or they may find ways to circumvent the physical security precautions. but a knowledgeable attacker with access can never be completely defeated. remote administration concerns. The details of the procedure varies between router models.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. Using this procedure. 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. One of the best additions to the security features of a computer network is to limit access. or electronic monitoring. system personnel. and configuration issues. There are things that can be done to make this more difficult. If remote administration is used to configure and control routers. 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]). These include physical access. as well as for your routers.1. consider the password recovery procedure for Cisco routers. should be located in a limited access area. Network infrastructure components. physical security mechanisms and policies must not make access too difficult for authorized personnel. In practice. This can be accomplished through the use of guards. “Step 1 Configure the router to boot up without reading the configuration memory (NVRAM).3).Advanced Security Services 4.1. Router Access Security This section discusses the various mechanisms used to protect the router itself. 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. To illustrate one reason why physical security is critical to overall router security. Network equipment. 4. Version 1.1. If possible. where applicable. This area should be under some sort of supervision 24 hours a day and 7 days a week. use encryption to protect the confidentiality and integrity of the remote administration connection (see Sections 5. like routers. especially routers and switches. an individual with physical access can gain full privileged (enable) access to a Cisco router without using a password. user account protection. but always includes the following basic steps. this area should only be accessible by personnel with administrative responsibilities for the router. then consider ways of protecting the machines used for remote administration. software protection. and the networks they use to communicate with the router. This is sometimes called the test system mode. It is always a matter of time and effort. Use access lists to limit remote administration access to hosts that enjoy reasonable physical security. When thinking about the security of your network it is important to consider these issues for all your systems. only slowed down.

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

authentication systems.9 was a mature version of Release 12). To check your IOS version. which can be used instead of numbered lists.2. A second problem is that the early versions of new IOS releases can be less robust than more mature. lines. the so-called ‘new model’.1.1. but not cutting-edge. or Vlan2. • access-list: There are two types of IP named access lists. access lists. Version 1. and much more. etc. For more details on IOS upgrades. log in and enter the command show version.1c 57 .1 was an early version of IOS Release 12. Newer versions of IOS fix bugs and vulnerabilities that existed in the older versions. • line (config-line) is used to set up the console port.6 for more details. Router Software Versions Cisco issues new IOS versions and upgrades fairly frequently. extended (config-ext-n) and standard (config-std-n). logging. and add new security features. including the ability to change the configuration.21).0. Cisco IOS releases. and logging router functions. Ethernet 0/1. The recommended minimum IOS release is IOS 12. To enter the global configuration mode (config) type the command configure terminal .5 and 8. later versions (i. Cisco IOS 11. This will allow others to find the bugs in the newer versions (and get them fixed). There are several configuration modes on a Cisco router. authorization. while 12. A good approach to this problem is to maintain operational routers with recent.0. EXEC mode gives limited access to the command set of the router. the system is in user mode also known as EXEC mode. commonly abbreviated “config t”. There are sub-modes which are used to configure specific settings for interfaces.3. auxiliary port and virtual terminal lines. and accounting (AAA).3. Privileged EXEC mode is sometimes called ‘enable mode’. making it a significant administrative burden to keep all the routers on a large network up to date. 12.1.1 and later offer a comprehensive model for authentication. • route (config-route) is where specific parameters can be set and modified for a selected routing protocol. In addition to the standard authentication.Advanced Security Services 4. 4. is reserved for the privileged EXEC mode. authorization.6 for a brief description and Section 4. see Sections 4. The list below describes some of the sub-modes. Router Configuration and Commands (IOS) After connecting to a router and initially logging in. • interface (config-if) is used to configure aspects of a particular interface like FastEthernet0.e. In the global configuration mode a wide variety of overall router features and settings can be changed: banners. Access-list mode is used for building named access lists. The recommended newest release would be the most recent “GD” version that is at least a month old (at the time of this writing. routing protocols.0. See Section 4. Typing the enable command at an EXEC mode prompt will give access to the privileged EXEC mode. Access to all the router commands. routing protocols.1. 12. Keep your IOS as up to date as is practical.

to define one loopback interface. we say that the service is bound to that interface. Passwords. Also. Commands to set source interface bindings are given with the discussion of each service in the rest of the guide. Cisco IOS routers have the ability to define internal virtual interfaces.255. 4. that connection must have some source address. Logins. simply assign it an IP address.5. When a router is configured to use the loopback interface for services. a statement about 58 Version 1.2. router network services that can be bound to the loopback interface should be. but routers also generate some network traffic. SNMP. called loopback interfaces.255. which includes a legal notice. one per line. Adopting this practice yields several benefits for the overall stability and security management of a network. For a border router. in configuring Cisco routers. because the address of the loopback interface is fixed.11. such as routing protocols. and TFTP.4. the loopback’s address usually should be in the range of the internal or DMZ network. It is considered best practice.Router Security Configuration Guide 4. In addition to physical interfaces. not the external network.255 Central(config-if)# end Central# In general. Central(config)# interface loopback0 Central(config-if)# description Main loopback interface Central(config-if)# ip address 14. should be set up on each operational router. Router Network Traffic and the Loopback Interface The pimary job of a router is forwarding traffic between networks. This can be problematic in several ways.1. it is possible to configure the security of other devices in the network more tightly. Routers and other network devices communicate using various management protocols. a statement that all use of the device must be authorized by the owning organization.1c . the loopback interface’s address does not appear in any route-based network maps. and designate it as the source interface for most traffic generated by the router itself. nor can it be part of the same network as any other interface. A legal notice usually includes a ‘no trespassing’ warning. typically a router will select a source address from one of the addresses bound to one of its network interfaces. End with CNTL/Z. NTP. consult [5]. hiding administrative aspects of your network from potential attackers is usually good practice. Central# config t Enter configuration commands. mainly because the source address for some services can vary. Note that the loopback address cannot be the same as the address of any other interface. and Accounts Logins and Banners A login banner.) To create a loopback interface.250 255.1. Privileges. It means that IP packets generated by the router will have the loopback interface’s address as their source address. For further discussion of loopback interfaces. (When a service is configured to use the loopback interface as its source. When the router initiates a network connection.

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

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. 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. 64.5 discusses configuration of the auxiliary port if it is required for a modem. Some IOS versions (including the versions designated “Enterprise”) may have 15.1c .3 for cryptographic mechanisms for protecting the remote administration connections. [Note: perform these commands only when connected to the aux or console port. To disable network virtual terminal connections to the router. one per line.1. It is important to know how many virtual terminals your IOS version has. or even more. Most versions of IOS have five virtual terminals. do not perform them while logged into the router via Telnet.1.] South# config t Enter configuration commands. see Section 4. Login on the virtual terminal lines should be disabled if remote administration is not absolutely necessary. If remote administration is necessary. 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. as shown in the example below. create an access list and apply it to the virtual terminal lines. and Sections 5. these connections are called virtual terminal lines. numbered 0 through 4.6 for details on configuring remote administration. or use the command transport input none. If the auxiliary port is required for a second local serial connection then configure it as shown below.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. and 60 Version 1.2 and 5.

If you do not know how many vtys your router supports.1c 61 . 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.Advanced Security Services to explicitly configure all of them securely. The transcript below shows how to delete the extra two vtys on the router South . Normally. (Note: on most IOS versions. 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. you can list them using the command show line vty in the manner shown below. in terms of security. By default Cisco provides EXEC (level 1) with a few commands which may. one per line. you cannot delete VTYs 0 through 4. then disable the extra ones.6 for more details). 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. Every IOS command is pre-assigned to either level 1 or level 15.) South# config t Enter configuration commands. The next example Version 1.simply delete 5. or delete them with the configuration mode command no line vty. you would configure all of the vtys on the router identically. Cisco IOS comes with 2 predefined user levels. End with CNTL/Z. and both 5 and 6 will disappear. make more sense being at a higher privilege level. If the router has more vtys than you need. two more than the default complement of five.

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

below). • Always include at least one of each of the following: lowercase letters. • The privileged EXEC secret password should not match any other user password or any other enable secret password. dates.3). Some of the more important rules are provided in the following list.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. Version 1. no user account should be created above privilege level 1.1 and earlier) • NTP authentication keys – for more information about NTP security. addresses. For access and logging reasons the login local method should be used. this will keep passersby from reading your passwords when they are displayed on your screen. digits.1 and earlier) – for more information about routing protocol authentication see Section 4. • Avoid more than 4 digits or same-case letters in a row. uppercase letters. Do not set any user or line password to the same value as any enable secret password. and special characters. • Enable service password-encryption. Then users with a need to work at higher levels would be given the higher privilege level password. proper names. • Avoid dictionary words.5. • Peer router authentication keys (in 12. and audit users (see Section 4. Good security practice dictates some other rules for passwords.1 and earlier) • TACACS+ keys (in 12. phone numbers.3. • RADIUS keys (in 12.1c 63 . But if the login local command is used to protect a line then the specified user name/password pair is used. • SNMP community strings – for more information about SNMP security see Section 4.4. Never set any of these secret values to the same string as any other password. • 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. see Section 4.5. AAA mechanisms may be used to authenticate. 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. • Make all passwords at least eight characters long. In addition to the above password access mechanisms.6 for details). authorize. But user accounts should be created for auditing purposes (see Accounts. • Be aware that there are some secret values that service passwordencryption does not protect.

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

Only if absolutely required should a modem be connected to the aux port as a backup or remote access method to the router. all connections to the router should require authentication (using individual user accounts) for access. the Internet) should not be used.5. anyone with a network sniffer and access to the right LAN segment can acquire the router account and password information. unless the traffic is adequately protected.3 describes how to use SSH for secure remote administration.1. This section will discuss remote internal only access without AAA. The Auxiliary Port As discussed in Section 4. Section 5.1 log Central(config)# access-list 99 permit 14. so it is necessary to apply access controls to the aux port.1c 65 . Attackers using simple war-dialing software will eventually find the modem. Remote access over untrusted networks (e.5 the aux port should be disabled. As discussed earlier. The five regimes listed above are listed in the order that best protects the router and allows for accounting of router activities. and set an exec session timeout. 12. such as IPSec or SSH. This can be accomplished by using login local (see next sub-section for example) or AAA (see Section 4. The security of remote administration can be enhanced by using a protocol that provides confidentiality and integrity assurances. When you use remote administration. and SSH should always be used instead of Telnet whenever possible. with or without AAA. create and apply an access list explicitly listing the hosts or networks from which remote administration will be permitted. For better security. To configure the vtys for remote access do the following: bind the telnet service to the loopback interface. Network Access Remote network connections use the VTY lines to connect to the router.6. This is why remote administration security issues center around protecting the paths which the session will use to access the router. Consult the Cisco IOS Dial Services guide [6] for information about connecting modems and configuring callback. A detailed discussion on setting up modems is beyond the scope of this document. Central(config)# ip telnet source-interface loopback0 Central(config)# access-list 99 permit 14. IOS callback features should be used. Setting up IPSec for remote administration is covered in Section 5.9.Advanced Security Services As discussed in Section 4.3T. Section 4.g.6 describes remote access with AAA.6).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 1.4 and later. Cisco has added support for the Secure Shell (SSH) protocol to many versions of IOS 12. because the user’s password will travel the network in clear text form.2. and nearly all IOS releases in and later. remote administration is inherently dangerous.

1c . Central(config)# service tcp-keepalives-in Central(config)# service tcp-keepalives-out Central(config)# exit Central# 4. enable TCP keepalive services. are less secure and should be avoided. Unless this capability is important for managing your network. and logging of user activities on a router. Finally. Authorization is the method used to describe what a user has the right to do once he has authenticated to the router.6 contains details on configuring AAA in an example network. These services will cause the router to generate periodic TCP keepalive messages. Central(config)# line vty 0 4 Central(config-line)# transport output none Central(config-line)# exit Lastly. 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). and Accounting (AAA) This is Cisco’s new access control facility for controlling access. privileges. Online editing provides for syntax checking but provides limited editing capability and no comments. Offline editing provides the ability to add comments. like rlogin and web. thus allowing it to detect and drop orphaned (broken) TCP connections to/from remote systems.1. the IP addresses which are allowed to connect must be on an internal or trusted network. 4.1. the transport input telnet command restricts the management interface to telnet only.3.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. Cisco IOS supports outgoing telnet as well as incoming. they can establish further telnet sessions from the router to other devices. Authentication is the mechanism for identifying users before allowing access to a network component. Authentication. once an administrator or attacker has gained telnet access via a VTY. Authorization. Section 4. Additionally. 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. Logistics for Configuration Loading and Maintenance There are two basic approaches for configuration loading and maintenance: online editing and offline editing. allows for the use 66 Version 1. For more details on access lists see Section 4. it should be disabled as shown below. They each have advantages and disadvantages.8. This is important because the other supported protocols. if you are going to permit remote administration via Telnet.7. Using this service does not remove the need for setting an exec-timeout time as recommended above.

then they will be stored in the clear and transferred in the clear. Because each Cisco IOS release changes the default values for some of the commands. using the terminal is the best choice for loading a new configuration. The example below shows how an encrypted enable secret setting would appear in an off-line configuration file. FTP (available for IOS 12. ! set the enable secret password using MD5 encryption enable secret 5 $1$fIFcs$D. 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.0 and newer). the system where the configuration files are stored should use the local operating system’s security mechanisms for restricting access to the files.8 Local and Remote Administration Section 4.1c 67 . the show run command will only show those configuration settings which are different from the IOS defaults.lgcsUnsgtLaWgskteq. The possible methods for transferring files to a router have increased with newer IOS releases. Instead. Third. maintain the configuration offline by writing it offline (see above). TFTP. Old and alternative configurations should be stored offline. First. When managing configuration files offline there are several security issues.3 recommends performing local administration. always use the startup configuration to avoid problems. tracking the configuration can become very difficult. use configuration management to track changes to your configurations. so that if there is a power failure or some other problem the router will restart with the correct configuration. Only save off the running configuration for an emergency. and then copying and pasting the encrypted string into your offline configuration file. rcp. if you set passwords in an offline configuration file. but provides no syntax checking. In this situation it is only necessary to manage the startup configuration since the running configuration is identical. because the saving will not include default values and after an IOS upgrade you may encounter unexpected configuration problems. You can obtain the encrypted string by setting the password manually on the router console. Only authorized router administrators should be given access to the files. then displaying the running configuration. and SCP (available in many releases 12. When saving and loading configurations. Second. with the configuration files offline the files must be transferred to the router in the relatively secure method. The primary mechanisms available are the console terminal. This is especially true for the enable secret password. Also. Cisco configuration save utilities will also not save default values.1 later that support SSH). it is best to type the passwords while on-line (using the console) and then copy the encrypted strings to the offline configuration. In this case. Only a few lines should be copied at a time so that you can determine that the entire Version 1. and guarantees all settings will be visible. But the offline method will leave passwords in the clear.Advanced Security Services of better editors. With the online editing. described below. It is also important to keep the running configuration and the startup configuration synchronized. The recommended approach is a hybrid of the two.1.

The following example shows how to save the startup configuration to a file.2. Central# copy running-config startup-config Central# config t Enter configuration commands. one per line. End with CNTL/Z.5516/1024 bytes] 5516 bytes copied in 4..364 secs Central# 68 Version 1. if you have one available.9. [Note: the default Windows NT 4. On Windows NT and 2000.9. use a better communication program. Erasing the nvram filesystem will remove all files! Continue? [confirm] y [OK] Erase of nvram: complete Loading /rtr-backup/central-config ! [OK .9. 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.Router Security Configuration Guide configuration file is transferred successfully. the tip command can be used.1/rtr-backup/central-config. otherwise use the interface closest to the FTP server. Central# config t Enter configuration commands. On Solaris. Set the source interface for FTP to the loopback interface if you have defined one.] 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 []? 14.352 secs (459 bytes/sec) Central# The next example demonstrates how to load a new configuration file into the startup configuration.0 serial communication program. 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 using the console connection or a telnet connection is the best choice for administration.2. On Linux. End with CNTL/Z.1 Source filename []? /rtr-backup/central-config Destination filename [startup-config]? Accessing ftp://14. Hyperterminal. one per line.2.1c . performs copy/paste very slowly.1 Destination filename [startup-config]? /rtr-backup/central-config Writing central-config !! 5516 bytes copied in 12. If remote administration is allowed and the IOS is newer then version 12.0 then use the FTP protocol to transfer the configuration files to and from the router. such as TeraTerm Pro. the minicom program is suitable for Cisco local console access.

available at: http://www. SCP should be used whenver possible. [4] “Password Usage” Federal Information Processing Standard Publication 112. modem. [2] Buckley. and more. 1999. [7] Akin.Advanced Security Services The other protocols. Cisco Press. Hardening Cisco Routers. Cisco IOS Essentials. and Wright. 1st Edition. Version 1.nist. 1999. This is the reference manual and guide for basic IOS configuration tasks. because it provides integrity and confidentiality protection. P. Cisco Press.5. National Institute of Standards and Technology. 1985. and Smith. Configuring Passwords and Privileges. vtys. J. especially for initial set-up of a new router. [5] Greene.0. References [1] Cisco IOS Release 12. all with fully worked-out examples. Cisco Press. An excellent introduction to basic usage and configuration of IOS routers. The sections about passwords and warning banners contain very good information. Relevant sections include: IOS User Interfaces and File Management.. Securing Cisco Routers: Step-by-Step. Cisco ISP Essentials. This is the reference manual and guide for serial line. See Section 4. A. and dial-in features. McGraw-Hill.9. ed. A very specific guide to configuring many IOS features securely.0 Security Configuration Guide.1.itl. This is the reference manual and guide for major security features in IOS 12. A pragmatic and detailed guide to securing Cisco routers. 4.5 for details on using TFTP if required. are less secure than FTP and should not be used for loading or saving router configurations.. J. 2002. It includes information about configuring logins. This detailed Cisco guide for Internet Service Providers includes extensive discussion of routing protocols (especially BGP). and an in-depth treatment of Unicast RPF.htm This federal standard includes some good guidelines on choosing passwords that are difficult to guess. 2002. O’Reilly & Associates. April 2002. Relevant sections include: Security Overview. [8] Stewart. [3] Albritton. B. J. Cisco Press. 1999. 2000.. [6] Cisco IOS Dial Services Configuration Guide. and Traffic Filtering and Firewalls. Cisco IOS 12. SANS Institute.1c 69 .0 Configuration Fundamentals. such as rcp and T.

Others are automatic processes and settings intended to support legacy or specialized configurations but which are detrimental to security.2: enabled 11. For example. chargen. As stated in Section 3. Required Services.2.3. etc. Unix user lookup service. but some other host is acting as the bootp server. most of the services listed below are not needed. Unauthorized persons don’t need to know this. etc. IP source routing should be disabled. 4. This is a legacy feature.3: disabled 11. or restricting access to particular network segments or sets of hosts.0. Turning off a network service on the router itself does not prevent it from supporting a network where that protocol is employed. Default 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. In this case.2. In many cases. 12. disable it explicitly. This list has been kept short by including only those services and features that are security-relevant and may need to be disabled. and Security Risks The table below lists some of the services offered on Cisco IOS 11. If a particular portion of a network needs a service but the rest does not. Standard TCP network services: echo. Some of these services can be restricted or disabled. IP source routing is a little-used feature of IP that can be utilized in network attacks. general security practice for routers should be to support only traffic and protocols the network needs.2: enabled Enabled Recommendation CDP is almost never needed. improving security without degrading the operational use of the router. disable it. and 7. disable it explicitly. Typical Services. 3. 70 Version 1. Standard UDP network services: echo. Cisco IOS supports turning a service off entirely. the router’s bootp server should be disabled. For example.Router Security Configuration Guide 4. Unless it is required for the network to operate.1c . then the restriction features should be employed to limit the scope of the service. 4. a router may support a network where the bootp protocol is employed.1. Router Network Service Security Cisco routers support a large number of network services at layers 2. and later versions. Some of these services are application layer protocols that allow users and host processes to connect to the router.3: disabled 11. discard. allows remote listing of logged in users. This is a legacy feature. disable it. 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.

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

CDP The Cisco Discovery Protocol is a proprietary protocol that Cisco devices use to identify each other on a LAN segment. To turn off CDP entirely. End with CNTL/Z. 13 .9. and how to disable the TCP and UDP small servers. The example below shows how to test whether the TCP small servers are running. In virtually all cases.2..2.2.250. use the commands shown below in global configuration mode.9. In the unlikely event that CDP is needed for part of a network.. and then disable it on each interface where it is not needed using the no cdp enable command in interface configuration mode. one per line.2. 13 .250.1c . and is considered deleterious to security. it can be enabled and disabled for each interface.3 and 12.2.250 daytime Trying 14.2. 2000 11:48:39-EDT [Connection to 14.9. April 3. To enable CDP use the cdp run command in global configuration mode. TCP and UDP Small Servers The TCP and UDP protocol standards include a recommended list of simple services that hosts should provide.2. It is useful only in specialized situations. % Connection refused by remote host Central# 72 Version 1. it is not necessary for routers to support these services. then tcp-small-servers are enabled Central# connect 14. Central# ! if connect succeeds. and they should be disabled.. How to Disable Unneeded Features and Services Each sub-section below describes how to disable or restrict particular services and features under Cisco IOS 11.250 closed by foreign host] Central# config t Enter configuration commands.9.250 daytime Trying 14. one per line.Router Security Configuration Guide 4. Open Monday. 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. Central(config)# no cdp run Central(config)# exit Central# show cdp % CDP is not enabled Central# End with CNTL/Z..9.

Version Security Services Finger Server The IOS finger server supports the Unix ‘finger’ protocol.9. On a Cisco router.9.9.250. and attacking a router..2..9..6 00:00:00 goldfish *131 vty 1 idle 00:00:00 central [Connection to 14. web-based remote administration should be avoided..9. While the web access features are fairly rudimentary on most Cisco router IOS releases. then it should be disabled as shown below. However. Open This is the CENTRAL router.250 closed by foreign host] Central# config t Enter configuration commands. Therefore. access restricted. users who are not authorized to log in to the router have no need to know who is logged in.2. If web-based remote administration is not needed.250 finger Trying 14.2. 79 . If web-based administration is examined and found necessary for network operations.2. then its use should be restricted as follows. Further.2. Typically. End with CNTL/Z. the show users command may be used to list the logged in users. web-based administration imposes the requirement that users log in at full (level 15) privilege. they are a viable mechanism for monitoring. Line User Host(s) Idle Location 130 vty 0 14. Central(config)# no ip finger Central(config)# no service finger Central(config)# exit Central# connect 14. one per line. 80 ..250 finger Trying 14. 79 .250. % Connection refused by remote host Central# End with CNTL/Z. it is important to note that both Telnet and web-based remote administration reveal critical passwords in the clear. The example below shows how to test and disable the finger server.9.250 www Trying 14.2.. Central(config)# no ip http server Central(config)# exit Central# connect 14. Web-based remote administration is useful primarily when intervening routers or firewalls prevent use of Telnet for that purpose.250. one per line. % Connection refused by remote host Central# HTTP Server Newer Cisco IOS releases support web-based remote administration using the HTTP protocol. configuring. which is used for querying a host about its logged in users.1c 73 .2. Central# connect 14. Central# config t Enter configuration commands.

0.2. Cisco IOS does not yet support the superior HTTP digest authentication standard). 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. 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. protect the HTTP traffic by setting up IPSec. Central(config)# ! Add web admin users. Cisco routers are capable of acting as bootp servers. • Create and apply an IP access list to limit access to the web server. To disable bootp service. (Some recent versions of IOS can support SSL (HTTPS) for protecting administrative HTTP traffic. If possible.2.2. Central(config)# no ip bootp server Central(config)# exit End with CNTL/Z. This guide does not provide instructions on configuring SSL.9. In practice. • Configure and enable syslog logging as described in Section 4.0 network and the host 14.18 log Central(config)# access-list 29 permit 14.1.5. one per line.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. bootp is very rarely used. primarily for other Cisco hardware.2.1c .0. Access lists are described in Section 4.18 only. and offers an attacker the ability to download a copy of a router’s IOS software.9.0 0. The example below illustrates each of these points. HTTPS is better than plain HTTP. one per line. as described in Section 5.2. The router’s web server will use HTTP basic authentication to demand a username and password (unfortunately. Central# config t Enter configuration commands. 74 Version 1. End with CNTL/Z. but IPSec is preferred.6. Administrators will be allowed to connect from the 14. Central# config t Enter configuration commands. consult the Cisco IOS documentation.6. AAA will give more control and better audit. as discussed in Section 4.3.2.) Bootp Server Bootp is a datagram protocol that is used by some hosts to load their operating system over the network.Router Security Configuration Guide • Set up usernames and passwords for all administrators. use the commands shown below. use AAA user access control as described in Section 4.6.

one per line. 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.1c 75 . Version 1. effectively extending a LAN at layer 2 across multiple segments. This feature is used in several kinds of attacks. Central(config)# no boot network Central(config)# no service config Central(config)# exit Central# End with CNTL/Z. A Cisco router can act as intermediary for ARP. ARP transactions are confined to a particular LAN segment. Cisco routers normally accept and process source routes. Explicitly disable loading the startup configuration from the network using the commands shown below. Unless a network depends on source routing. Normally. Loading from the network is not secure. Because it breaks the LAN security perimeter. Disable it on each interface where it is not needed. it should be disabled on all the net’s routers. and should be considered only on a network that is wholly trusted (e. and only when absolutely necessary to support legacy network architectures. responding to ARP queries on selected interfaces and thus enabling transparent access between multiple LAN segments. using the interface configuration command no ip proxy-arp .25.25 links.Advanced Security Services Configuration Auto-Loading Cisco routers are capable of loading their startup configuration from local memory or from the network. Cisco routers perform proxy ARP by default on all IP interfaces. This service is called proxy ARP. Disable it from global configuration mode as shown below.g. PAD Service The packet assembler/disassembler (PAD) service supports X. even on interfaces that are currently idle. Central(config)# no service pad Central(config)# IP Source Routing Source routing is a feature of IP whereby individual packets can specify routes. a standalone lab network). Central# config t Enter configuration commands. but it is not needed unless your router is using X. proxy ARP should be used only between two LAN segments at the same trust level. The example below shows how to disable IP source routing. The example below shows how to disable proxy ARP on four Ethernet interfaces. This service is on by default.

Three ICMP messages are commonly used by attackers for network mapping and diagnosis: ‘Host unreachable’. The example below shows how to turn them off for an interface. 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.15. Automatic generation of these messages should be disabled on all interfaces.1. 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. 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.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. IP Unreachables. Redirects. End with CNTL/Z. Mask Replies The Internet Control Message Protocol (ICMP) supports IP traffic by relaying information about paths.250 YES NVRAM up up Ethernet0/1 14.Router Security Configuration Guide Central# show ip interface brief Interface IP-Address OK? Method Status Protocol Ethernet0/0 14. especially interfaces that are connected to untrusted networks. and network conditions. one per line. Central# config t Enter configuration commands. one per line. Cisco routers automatically send ICMP messages under a wide variety of conditions. and the default Cisco IOS configuration is to reject directed broadcasts. routes. End with CNTL/Z.2. This technique was used in some old denialof-service attacks. and ‘Mask Reply’.1c . ‘Redirect’.9.

10. It is enabled by default on Ethernet interfaces in some versions of IOS. 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. Disable it on each Ethernet interface as shown below.Advanced Security Services MOP The Maintenance Operations Protocol (MOP) was used for system utility services in the DECnet protocol suite. • Disable SNMP system shutdown and trap features.5. as described in Section 4. End with CNTL/Z.1. SNMP Services The Simple Network Management Protocol (SNMP) is the standard Internet protocol for automated remote monitoring and administration. with different security properties. then disable NTP as shown below. all SNMP facilities on all routers should be disabled using these steps: • Explicitly unset (erase) all existing community strings. If possible. one per line. • Disable SNMP system processing.1. In the absence of a deployed SNMP scheme. The example below shows how to disable SNMP by implementing these recommendations. use an access list.1c 77 .250 YES NVRAM up up North# config t Enter configuration commands.20 YES NVRAM up up Ethernet1/0 14.3. If a network has a deployed SNMP infrastructure in place for administration. as discussed in Section 4. then all routers on that network should be configured to securely participate in it. If NTP services are not available on the network. North# show ip interface brief Interface IP-Address OK? Method Status Protocol Ethernet0/0 14. To reject all NTP messages at a particular interface. configure all routers as part of an NTP hierarchy. It starts with listing the current configuration to find the SNMP Version 1. 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.2. There are several different versions of SNMP.

one per line. If one or more name servers are available on the network. then explicitly set the name server addresses using the global configuration command ip name-server addresses. By default. you must simply list the entire configuration and inspect it visually)..255. The configuration listing is often quite long. IOS sends DNS name queries to the broadcast address 255.5. using the 78 Version 1. shuts down all SNMP processing on the router.0 and earlier. no snmp-server. is to list the established SNMP community strings and explictly unset them as shown above.1c . and you want to be able to use names in IOS commands. and will remain so. 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. turn off DNS name resolution as shown below. Central# show running-config | include snmp Building configuration. some SNMP configuration statements will not appear in any listing of the running configuration. For information on setting up and using SNMP securely.. see Section 4. DNS name resolution should be enabled on a router only if one or more trustworthy DNS servers are available. End with CNTL/Z.255. snmp-server community public RO snmp-server community admin RW Central# Central# config t Enter configuration commands. one per line. In general. but they can still be there! The safest way to ensure that SNMP is really unavailable to an attacker. note that SNMP must be enabled in order for the SNMP community strings to appear in the configuration listing. It is also a very good idea to give the router a name. North(config)# no ip domain-lookup North(config)# end End with CNTL/Z. Router Name and DNS Name Resolution Cisco IOS supports looking up host names with DNS. North# config t Enter configuration commands.255.3. When SNMP processing is shut down.Router Security Configuration Guide community strings. so you may want to use IOS output filtering to display only the lines related to SNMP (under IOS 12. If you do not want your router to send queries.

Central# config t Enter configuration commands.2 14. To set a domain name. This sample is formatted as it would appear in a configuration text file stored on a host for download to the router Central.testnet. Setting a domain name is also necessary for using SSH (see Section Security Services command hostname. 4.9. router# config t Enter configuration commands. For more information about NTP and SNMP security configuration.3).2.4. and enforces the need for router administration privileges when adding new network connections to a router. one per line.1 Central(config)# ip domain-lookup Central(config)# You can also set a default DNS domain name. Configuration Example The configuration listing below shows the configuration commands for disabling typical unneeded services. ! ----. as described above. use the config command ip domain-name domain as shown below. use the command shutdown in interface configuration mode. Central(config)# ! full name of this router: Central.5. and set up a main and backup DNS server address for the router Central. one per line. the name you give to the router will appear in the prompt. Central(config)# interface eth0/3 Central(config-if)# shutdown Central(config-if)# end Central# End with CNTL/Z. see Section 4. This helps discourage unauthorized use of extra interfaces. Disable Unused Interfaces It is a good idea to explicitly shut down (disable) unused interfaces on your router. End with CNTL/Z. To disable an Central(config)# ip domain-name Central(config)# end Central# 4.1c 79 .1.3. The example below shows how to set the router name. which will be used as part of the fullyqualified host name of the router and any unqualified name lookups. router(config)# hostname Central Central(config)# ip name-server 14.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.

9.Router Security Configuration Guide no no no no service pad ip http server ip name-server ip domain-lookup ! ----.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.Per-interface services section interface eth 0/0 description Outside interface to 14.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 1.2.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 ! ----.1c .1.0.

Cisco Press. September Available from Cisco’s web site.5. [5] Cisco IOS Network Protocols Configuration Guide. available at: http://www. [2] “Cisco IOS Version 12. New York. McGraw-Hill. Version 1. 81 . Cisco Systems. 1999. Short article with some good advice on features to turn off. includes common-sense measures to take on routers running IOS 11.0 Security Configuration. though.phrack. The sections on “Performing Basic System Management” and “Monitoring the Router and Network” include valuable advice on how to configure basic features and services. [6] Held.0 Configuration Fundamentals.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. [3] “Increasing Security on IP Networks. Cisco Press.2. Part 1. 9 Issue 55.3. Cisco Security Architectures. Can’t seem to find it on the web anymore. [4] Buckley. 1998.” . available under: http://www. 1999.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.” National Y2K Information Coordination Very helpful article from Cisco. Cisco Internetworking Case Studies. The section on “IP Addressing and Services” includes information about several of the IP services described in this section. K. 1999. plus excellent coverage of access lists. September 1999. Vol. Good overview of Cisco router and TCP/IP architecture. A.” Phrack Magazine. and Hundley. Cisco IOS 12. B. “Building Bastion Routers Using Cisco IOS. References [1] Eldridge.

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

source is the IP address of the network or host from which the packet is being sent. Version 1. Concepts Access lists on Cisco routers provide packet selection and filtering capabilities. there are two types of access lists available: standard and extended. Access Control Lists. The condition field. For IP traffic. deny denies access if the condition is matched. source or destination TCP/UDP ports. specifies which packets match the rule. 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 99.3. for restricting access to services on the router itself. Conditions typically involve protocol information and addresses. Extended access lists can permit or deny packets based on their protocols. or ICMP or IGMP message types. but do not involve application-level information. Standard access lists only allow source IP address filtering. which is different for each kind of access list. and what kind of access list it is. Extended access lists also support selective logging.1. access-list list-number {deny | permit} condition The access list number tells Cisco IOS which access list the rule should be a part of. routing protocols.3. and many router features. IPSec. Secure configuration of Cisco routers makes very heavy use of access lists.1c 83 . Only standard IP access lists can be applied to SNMP. 4. 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). and for other packet identification tasks. permit permits access if the condition is matched. source or destination IP addresses. Filtering. Syntax The basic structure for an access list rule is shown below. An access list consists of one or more rules. with some extensive examples. source-wildcard is the wildcard bits to be applied to the source. and for filtering traffic passing through the router. This section gives a moderately detailed description of access list syntax. Both standard and extended IP access lists can be applied to router interfaces.Advanced Security Services 4. vty lines (for remote access).

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

In each access list there must be at least one permit statement.0.1.Advanced Security Services keyword.0. and any TCP traffic from the Internet to all hosts in the 14.1.2 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.2.255 North(config-ext-nacl)# deny ip any any log North(config-ext-nacl)# exit General Recommendations Refer to the two tables in Section 3.0.255. Syntax Examples The example below shows how to create a small extended IP access list that permits DNS traffic to the address 14.0. using the router East as an example.1c 85 . all TCP traffic. The name cannot contain spaces or punctuation and must begin with an alphabetic character. name is the name of the access list. they should not be allowed across the router in either direction or to the router. vty lines and routing protocols.1.2. For any connection that requires two-way interaction (e.1.0/16 network. extended specifies an extended IP access list. It is possible however to apply two access lists (one for each direction) for router interfaces.2 eq 53 North(config-ext-nacl)# deny udp any any log North(config-ext-nacl)# permit tcp any 14.0 0. North(config)# ip access-list extended border-filter-14 North(config-ext-nacl)# permit udp any host 14.0. The diagram below shows how access lists work when applied to router interfaces.255 any log The example below shows the same list as a named IP access list.2 eq 53 any log 14.1. an access list with no permit statements will block all network traffic wherever it is applied.1..1. ip access-list {standard | extended} name where standard specifies a standard IP access list.1.1.g.0 0. 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. Version 1.255. some UDP traffic) the access list will only affect approximately half the packets. The second table lists those services on the internal network or on the router that should not be accessible by external clients. Otherwise. Note that an access list is applied to packets traveling in one direction only. The first table lists those services that should be completely blocked at the router. After the permit or deny keyword the syntax is the same as defined above for either the standard list or the extended list.

This will guarantee that the router will log the values for the source and destination ports for TCP and UDP traffic.0 0.0.0 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. If you might apply the same access list to more than one interface.6.6.20 East Eth1 14.Router Security Configuration Guide 14.0/24 Interface Eth0 Interface Eth1 permit 14. When the access list is ready you can cut and paste the access list via a connection to the router.1.0. then use the qualifier log-input instead of log. Logs of denied packets can be useful for detection and analysis of probes and attacks against a network.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.250 14. These statements include the entire port ranges for TCP and UDP explicitly. 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.0. East(config)# access-list 102 permit ip 14.6.0 network Inbound Access List deny permit Inbound Access List deny Eth0 14. it is best to build it offline on a separate computer. as shown in the example below.1c .6. Log messages generated by access lists are at log level 6 ‘Informational’.1.2. Thus. Access list log messages always include the access list number. This feature will provide valuable information about what types of packets are being denied. due to limited editing capability on the Cisco router. you cannot easily modify access lists.1.2. Since the original access list is 86 Version 1.1. whenever you need to change an access list. which is usually sufficient to identify the provenance of the traffic.

East(config)# no access-list 100 Be careful when clearing and rebuilding an access list that is in use. traffic load. SNMP should be used only on internal or protected networks.6.2. 4.18 to connect to the router East via Telnet. The list denies all other connections.2.2 and 4.6 any eq 23 log East(config)# access-list 105 permit tcp host 87 . Filtering Traffic to the Router Itself Access lists are used in a variety of ways to control access to services on the router itself. Most of the time. Typically. When SNMP service is enabled on a router.6 to gather SNMP information from the router. While it is possible to incorporate access controls for these services into the access lists placed on interfaces.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. It also logs all successful and unsuccessful connections.3. see Section 4. your changes will take place instantly.2.6. This access list allows the host with IP address It is safest to build an access list completely first. Remote Login (Telnet) Service There are a number of methods to filter access to the router itself: vty lines. Below is an example of how to clear an access list. Versions 1 and 2 of SNMP are not considered secure due to the lack of strong authentication. Thus.2. The following example shows the configuration of a standard IP access list that is applied to a snmp server. possibly dropping traffic or even leaving the router vulnerable. see Sections 4. SNMP servers and routing protocols. East(config)# access-list 75 permit host 14. then apply it to the interface or service where you need it. network management tools can use it to gather information about the router configuration. and how to disable unneeded ones. Version 1.6 and 14.6.6. route table. For more information about services on the router. East(config)# access-list 105 permit tcp host 14. The following example shows the configuration of an extended IP access list that is applied to the vty lines. and more efficient to use the specialized facilities that IOS makes available to apply access controls directly to the services themselves. it is typically easier.Advanced Security Services still on the router.6 East(config)# access-list 75 deny any log East(config)# snmp-server community N3T-manag3m3nt ro 75 For more information about SNMP configuration. you must purge it before adding the updated access list. The list denies all other connections. and more. a router administrator telnets to one of the vty lines. This simple IP access list allows the hosts with IP addresses 14. more reliable.2. The vty lines are used for remote access to 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


11 host 0.255 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 log access-list 102 permit icmp 14.0 any parameter-problem access-list 102 permit icmp 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 0.12 host 0.255 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 gt 1023 any lt 1024 access-list 102 permit udp any packet-too-big access-list 102 permit icmp 14.0 eq 23 log access-list 150 permit tcp host 14.6.10 host 0.250 host 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 14.0 0.0 eq 23 log access-list 150 permit tcp host host 14.0 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 1.2.255 any echo access-list 102 permit icmp 14.255 access-list 100 permit ospf 14.1c . 0.0 0.0.0 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 gt 1023 any eq 53 access-list 102 permit udp 14.0.0 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 14.0 0.0.

To enable turbo access lists on a router. 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. you can view statistics about them using the command show access-list compiled. Once you have enabled turbo access lists. In the latter case. CAR Command Syntax Configuring CAR requires you to apply rate limiting rules to each interface where you enforce constraints on traffic or bandwidth usage. and later. (If your IOS does not support compiled access lists. This short section gives an overview of CAR.) Once this facility is enabled.3. use the command show interface rate-limit. By allocating a specific amount of bandwidth to defined traffic aggregates. data passing through the router can be manipulated to preserve fragile traffic. and a few simple examples. ordered set of rules for the in-bound (receiving) and out-bound (sending) directions. somewhat simplified. For more information on CAR commands. Each interface can have a separate. Turbo Access Control Lists Some Cisco router models support compiled access control lists. 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. consult the “IOS Quality of Service Solutions Command Reference” section of the IOS documentation. To view the CAR rules on all the interfaces. Using compiled access control lists can greatly reduce the performance impact of long lists. The output of the command will show both the rules and some traffic statistics about the rate limiting. then compiled ACLs can give improved performance. as shown in the examples below. the most important task that CAR can perform is to mitigate the paralyzing effects of DoS attacks and flash crowds. the command will generate a harmless error message. but to be ready to apply them quickly when you detect an attack in progress.3.1(6). however. enter it again adding the keyword no to the front. 4. and limit spoofed traffic. Version 1. You can use CAR to reserve a portion of a link’s bandwidth for vital traffic. IOS will automatically compile all suitable access lists into fast lookup tables while preserving their matching semantics. A sample of the output is included in the first example below.5. it may not be necessary to keep CAR rules in place at all times. To remove a rule. in IOS 12. or to limit the amount of bandwidth consumed by a particular kind of attack.1c 95 . use the configuration mode command access-list compiled. simply type the rule in interface configuration mode. The general syntax for a CAR rule is shown below.6.Advanced Security Services 4. If you use access lists with six or more rules on high-speed interfaces. called “Turbo ACLs”. eliminate excessive traffic.

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

action: continue last packet: 2668ms ago. 225000 extended limit conformed 346 packets.1c 97 . 112000 limit. 2500 limit. The rest of the link’s bandwidth will be usable by excess SMTP traffic and all other IP traffic. current burst: 0 bytes last cleared 00:02:32 ago. exceeded 0 bps matches: access-group 131 params: 16000 bps. 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. 24990 bytes. action: continue exceeded 255 packets. exceeded 0 bps North# In this second example. 25000 limit. action: drop last packet: 7120ms ago. CAR is being used to throttle a TCP SYN flood attack. action: transmit exceeded 0 packets. conformed 0 bps. exceeded 990 bps matches: all traffic params: 9000000 bps. 0 bytes. CAR is used to reserve 10% of a 10Mb Ethernet link for vital outgoing SMTP traffic. 50000 extended limit conformed 12 packets. conformed 2000 bps. 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. current burst: 0 bytes last cleared 00:01:40 ago. 27074 bytes. 2500 extended limit conformed 130 packets. you might want to impose both outbound and inbound rate limiting to protect the vital SMTP traffic.Advanced Security Services CAR Action Syntax set-prec-transmit prec Action Performed Set the IP precedence to prec and transmit or forward the packet. In practice. 12740 bytes. 11699 bytes. current burst: 2434 bytes last cleared 00:02:04 ago. CAR Examples In the first example. conformed 0 bps. action: drop last packet: 7140ms ago. and to limit outgoing ICMP ‘ping’ traffic to less than 1% of the link. 0 bytes. action: transmit exceeded 0 packets.

Take care when defining and applying CPP policy -. To implement a CPP policy. or rate limited. In this case. CPP protects the central processor via policies that filter or rate limit traffic directed to the processor. The forwarding plane (also called the “data” plane) forwards user data packets through the router.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 “critical.” “malicious. 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. discarded. consult [10]. all traffic destined for the control plane of a router must be categorized into network administrator-defined groups or classes (e. and management.1c . identify the classes you wish to handle. Once you have defined your classes. The management plane consists of traffic for configuring and monitoring router operations. signaling and link management protocols.g. Detailed information about CPP may be found in a Cisco white paper [12]. setting up control plane policing on IOS requires four steps. In the map. Control Plane Policing (CPP) Conceptually. The control plane consists of the routing.” “normal. 4.) 2. (If you have a ‘default’ class. Define a named class map for each of the access lists you created in step 1. and rough traffic rate limits for each of them. Then service policies should be created and applied that cause traffic classes destined for the route processor to be accepted.3. control. Timely and reliable operation of the management and control planes are essential for maintaining the flow of traffic through the forwarding is easy to accidentally restrict the wrong traffic and disrupt management or control plane services.” and “default” classes). use the class map-name command to define rate-limiting policy for each 98 Version 1. legitimate traffic would also be affected. For another example of using CAR to combat a DoS attack. Before attempting to configure CPP. 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). Create a policy map using the policy-map command. do not create an access list for it. 1. Create access lists that match (permit) the traffic from members of each class. router operations can be abstracted into three planes: forwarding.7. using the class-map command. 3.

255 any North(config)# access-list 151 permit ip 14.1. 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.Advanced Security Services named class. Define a default rate-limiting policy using the command class class-default.2. Traffic from hosts in the trusted class will have no rate limits. North(config)# ! define ACL for CPP trusted hosts North(config)# access-list 151 permit ip 14. The example below shows how to configure CPP with three different classes: a trusted class for internal and specific external hosts. Traffic from all other hosts will be rate-limited to 150 packets per second. End with CNTL/Z.3. Apply your policy map to the control plane using the commands control-plane and service-policy.255. and the bandwidth required to maintain router operations.255.12.255 any North(config)# access-list 151 permit ip host 7.1c 99 .2.4 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. consider the bandwidth from possibly hostile sites. and a default class for all other addresses.1. a malicious class for a known hostile host.0. Traffic from the malicious host will be dropped entirely. 4.0 0.0 0. When planning your CPP rate limits. North# config t Enter configuration commands. any North(config)# ! define ACL for known hostile host North(config)# access-list 152 permit ip host 1. one per line.

SMTP. FTP. McGraw-Hill. [2] Karrenberg..0 . 1999. Cisco Press. This RFC describes the IP address allocation for private intranets.8.0.0 .0.. and a good primer on access lists [5] Cisco IOS Release 12. Elizabeth D.255. RFC 1918. and Hundley.1c . RFC 2827.. Cisco Security Architectures. Brent and Zwicky.168. D.0.255. To remove a CPP policy. use the command show policy-map control-plane.g. and dynamic access lists. [4] Held. G.. K. and Hundley.Router Security Configuration Guide To view the current CPP policy and traffic statistics. It includes information on TCP Intercept.255. Moskowitz. K.31. O’Reilly Associates. This book offers detailed information about access control lists and many examples of list syntax and usage.255. B. Telnet. 1995.0 . [6] Ferguson.172.192.0. “Network Ingress Filtering: Defeating Denial of Service Attacks which employ IP Source Address Spoofing”. February 1996.3. etc. References [1] Chapman.0. Cisco Access List Field Guide. This text provides valuable information on how to packet filter many of the commonly used services. P. D. Building Internet Firewalls. This is the reference manual and guide for major security features in IOS 12. and Senie. 100 Version 1. “Address Allocation for Private Internets”.0 Security Configuration Guide. This Internet ‘Best Current Practice’ RFC gives a good overview of source address filtering. and Rekhter.. 2000. Y. reflexive access lists. use the command no service-policy command as shown below.255.168. e. and 192. D. 1999 This book includes a good introduction to router security. North(config)# control-plane North(config-cp)# no service-policy input cpp-policy North(config-cp)# end North# 4.255. 1999. McGraw-Hill.. G. [3] Held. 172.10. The Internet Assigned Numbers Authority has reserved the following three blocks of the IP address space for private intranets: 10.16..255.

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

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

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

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

but is unsuitable for large networks because it increases administrative overhead and requires administrative response to any failures. Anyone on a LAN segment can modify an entry in the ARP cache of Version 1. Authenticated router updates ensure that the update messages came from legitimate sources. This can be accomplished in many different ways. There are two basic approaches available for protecting route table integrity: 1. is the protocol used to map IP addresses to a particular MAC or Ethernet address. and recover from them quickly. an attacker who sends false routing update packets to an unprotected router can easily corrupt its route table. ARP and LANs Address Resolution Protocol. preventing router update messages from being sent or received will result in bringing down parts of a network. This section focuses on security issues related to routing and routing protocols. network administrators can deter attacks based on unauthorized routing changes. By doing this. because ARP offers no security. For example.2. 2. Routing security should be a top priority for network administrators who want to: • prevent unauthorized access to resources on the network. Use only static routes – This may work in small networks. 4.1c 105 . However. • protect mission information from unauthorized exposure and modification. An unprotected router or routing domain makes an easy target for any network-savvy adversary. For a detailed description of Proxy ARP. The fundamental security weakness of ARP is that it was not designed to use any form of authentication. bogus messages are automatically discarded. Another form of attack an adversary might attempt against a router is a denial of service attack. For example. or ARP. 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.Advanced Security Services could an adversary do to our network?” Section 3 presents some motivations for overall router security.4. routers need rapid convergence and backup routes. A detailed analysis of routing protocol threats and countermeasures may be found in a Cisco SAFE white paper [45]. Authenticate route table updates – By using routing protocols with authentication. the attacker can re-route network traffic in whatever manner he desires. • prevent network failures and interruptions in service. consult RFC 1027. neither does Proxy ARP. ARP is described in more detail in RFC 826 and Parkhurst [2]. To resist denial of service attacks. The key to preventing such an attack is to protect the route tables from unauthorized and malicious changes.

it is susceptible to bad or malicious routes. Such a compromise might lead to re-routing of traffic. because it is a good illustration of the basic principle. routers connected to the same network segment all use a shared secret key. 4. but instead uses Proxy ARP to handle the routing. The main focus of this section is using peer router authentication with interior gateway protocols.5 gives some security guidance for one exterior gateway protocol.1c .4. static routes. or corrupted routing updates that would compromise the security or availability of a network. Authentication ensures that the receiving router incorporates into its tables only the route information that the trusted sending router really intended to send. and routing protocols This section describes how to protect routers from some common routing hazards. The following example illustrates how to do just that. This sub-section describes the implementation of router neighbor authentication in OSPF. In any case. Central# config t Enter configuration commands. BGP-4. Each sending router then uses this key to ‘sign’ each route table update message. When a sending router builds an OSPF packet.4. The receiving router checks the shared secret to determine whether the message should be accepted. Section 4. when applied correctly. 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. it signs the 106 Version 1. malicious. a denial of service. can prevent many routing attacks.3. Router Neighbor Authentication The primary purpose of router neighbor authentication is to protect the integrity of a routing domain. Route tables. or simply giving access to certain packets of data to an unauthorized person. Each router accomplishes authentication by the possession of an authentication key. authentication occurs when two neighboring routers exchange routing information.Router Security Configuration Guide a router that serves the segment. OSPF Authentication Router neighbor authentication is a mechanism that. OSPF uses two types of neighbor authentication: plaintext and message digest (MD5). In this case. Therefore. It prevents a legitimate router from accepting and then employing unauthorized. if a host on the network does not use default gateways. Proxy ARP is generally not used anymore. That is. and it should be disabled. authentication in RIP version 2 and EIGRP work in a similar fashion. Plaintext authentication uses a shared secret key known to all the routers on the network segment. one per line.

Advanced Security Services packet by placing the key as plaintext in the OSPF header. r0utes-4-all. Figure 4-3 shows our example network with its routing protocols. Internet eth0/0 North eth0/1 If the keys match.1.2.9. they can pose as a trusted router. The second.9. with a Key ID of 1.64/24 South eth0/1 14. is message digest authentication.250/16 14. Using this method reveals the secret key to any attacker using a network sniffer on the right LAN segments.1. then the router accepts the packet.64/24 14.0.0/24 eth0/0 14.1.0/24 Central eth0/1 Autonomous System Border Router (ASBR) 14.1. and Central all share the same secret key.0/16 eth1 eth0 14.15.250/24 eth0/0 14. whose cryptographic Version 1.250/24 RIP Second Floor 14.6. Each of these routers authenticates to each other using the MD5 message digest authentication method.2.1c 107 .20/16 East 14.10.250/16 OSPF Area 0 Facility Network 14. the router rejects the packet.2. Otherwise. routers North. This method does not provide much security because the key is in plaintext in the packet.2.10.0/24 Figure 4-3: An Example Routing Architecture In this example.2. Once an attacker captures the key. East.1. The receiving router then compares the received key against the key in memory.2. and more secure method.

108 Version 1. It then picks a primary key to use on the network segment. The sequence number is incremented with every new packet. is written over the secret that was appended to the packet. Figure 4-4 shows how East authenticates to North. MD5. both header and body. If the regenerated hash matches the hash that was sent from East. East first builds an OSPF packet. The corresponding Key ID. The router then uses its own key to regenerate the hash on the received packet in the same manner as the sending router. looks at the Key ID to determine which key was used to generate the hash. then the North trusts the packet. In this case. East also places a 32-bit sequence number in the header. East runs the cryptographic hash algorithm. This sequence number protects against replay attacks so that no two OSPF packets will have the same hash value. use the superior MD5 method. or signature. North. is placed in the header. the secret key is appended to the packet. it rejects the packet as being invalid.1c . 1. The receiving router. Finally. against the OSPF packet. Otherwise. 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. 16 bytes. the key is r0utes-4-all. The output.Router Security Configuration Guide authentication type is denoted by a value of 2.

RIP MD5 Authentication The example below illustrates an example of setting up MD5 for RIP router neighbor authentication.0.255.255. The neighboring RIP routers use shared secret keys.0.Advanced Security Services OSPF MD5 Authentication The example below illustrates an example of setting up MD5 for OSPF router neighbor authentication.0. The example transcripts below show routers North and East receiving the key r0utes-4-all. all the routers participating in a given network should be configured in the same way.1.0 0.1c 109 . although the IOS commands are somewhat different. End with CNTL/Z. one per line. contained in their respective Version 1.6.255 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. RIP Plaintext Authentication This method is not recommended. North# config t Enter configuration commands. North(config)# router ospf 1 North(config-router)# network 14. below. The receiving router then uses the shared secret to check the hash and determine whether the message should be accepted. using the same key. one per line. Central and South. In practice. East(config)# router ospf 1 East(config-router)# area 0 authentication message-digest East(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.1.0. Each sending router uses these keys to generate the cryptographic hash incorporated into each RIP update message. End with CNTL/Z. use the superior MD5 method. RIP’s method of authentication is very similar to that of OSPF.0 0. Using the example network shown in Figure 4-1.0.0. router Central would also have to be configured with MD5 authentication and the same shared key as shown below.0 0. The example transcripts below show routers from Figure 4-3.255 area 0 East(config-router)# network 14. receiving the key my-supersecret-key.2.

each neighboring router must have a shared secret key. 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. The example below shows how to enable version 2 of RIP. However. The router examines the key numbers in order from lowest to highest. Central(config)# router rip Central(config-router)# version 2 Central(config-router)# network 14. This was a feature that was included in RIP version 2. Central and South have key chains named CENTRAL-KC and SOUTH-KC. Both key chains share the keys my-supersecret-key and my-othersecretkey. all the routes connected to a given network must be configured in the same way. Multiple keys with different lifetimes can exist. Central# config t Enter configuration commands.1c .0.0. and uses the first valid key that is encountered. RIP manages authentication keys by the use of key chains. both routers will only use the first valid key. That is. End with CNTL/Z. only one authentication packet is sent.Router Security Configuration Guide key chains. In the example below. A key chain is a container that holds multiple keys with the associated key IDs and key lifetimes. one per line. Prior to enabling RIP MD5 authentication. one per line. Each RIP router must first be configured to use version 2 in order to enable authentication during routing updates. all of them must possess the same shared key(s). 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 other key is usually used when migrating to different keys. However.0 Central(config-router)# end Central# South# config t Enter configuration commands. 110 Version 1. End with CNTL/Z. one per line. End with CNTL/Z. In practice. one per line. Central# config t Enter configuration commands.

0. Finally.Advanced Security Services South(config)# router South(config-router)# South(config-router)# South(config-router)# South# rip version 2 network 14. the key chain is defined within key chain configuration mode consisting of a key name. End with CNTL/Z. 4. key number. key number. The key chain name is locally significant and neighboring routers do not have to be configured with the same name. Router North’s key chain is defined as northkc and router East’s key chain is named eastkc. In this example.1c 111 .0 end Finally. End with CNTL/Z. one per line.0/16 network. and key string to be used. 2. This insures the integrity of routing messages accepted from neighboring routers. Router North has associated key number 1 with the Version 1.0. 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. Proceeding into the interface configuration mode. Initially. Central# config t Enter configuration commands. Authentication for RIP is enabled on the interfaces. Specify the key chain. The example below details the steps necessary to configure MD5 authentication on two EIGRP peers. North and East. 3. Enable authentication for EIGRP messages. 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. one per line. EIGRP is configured on both routers for the 14. MD5 authentication is enabled within autonomous system 100 and linked to a particular key chain. and key string.1. To configure EIGRP authentication: 1. Central will be using the key chain CENTRAL-KC that was created earlier and the MD5 method of authentication. the example below shows how to enable authentication for RIP.0. Configure key management (optional).

0.2.0 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.1.0. The router chooses the first valid key while examining the key numbers from lowest to highest. 112 Version 1.0. but only one authentication packet is sent. 2003 until January 1.1. East(config)# router eigrp 100 East(config-router)# network 14.255.0 255. Key management is optionally configured with the acceptlifetime and send-lifetime commands. one per line. the key-string “secret-key” is associated with key number 1. the routers will send updates authenticated with the key ‘my-secret-key’ from October 1. North(config)# router eigrp 100 North(config-router)# network 14. In this case.255. The examples below show how to configure EIGRP authentication and keys.1c .255.0. North# config t Enter configuration commands.0 255. 2004.0 East(config-router)# network 14. 2004. one per line.End with CNTL/Z. End with CNTL/Z.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.6.Router Security Configuration Guide key-string ‘secret-key’. In the example above.0 255.255. it will accept updates with that key until January 7. Multiple keys and key-strings can be configured on a router.

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

0001.1.1. one per line.e.1.0001.00 14. End with CNTL/Z.0000.1c .0000. For example.250 255. one per line.1. North(config)# router isis secure-network North(config-router)# net 49.0000. 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.250/24 eth0/0 14.0000.1.0001. CNSP. LAN Level 1: Hello packets).2.00 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 eth1 eth0 IS-IS Area 49.0001.255.15.20/16 East 14.0/24 Central Figure 4-5: An Example Routing Architecture Next.Router Security Configuration Guide East# config t Enter configuration commands.0001.0.0/16 NET = 49. and PSNP packets): North# config t Enter configuration commands. Also notice that IS-IS HMAC-MD5 authentication is applied to the IS-IS instance to authenticate routing database updates (i.6. Level 1: LSP. ensure the IS-IS routing protocol process is running on the routers. 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. 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 = 49.0 North(config-if)# ip router isis secure-network 114 Version 1.0003.0000. Then select the IS-IS PDU types to authenticate and enable IS-IS HMAC-MD5 authentication on the routers.250/16 14.1. The following scripts implement this security policy.0001 Facility Network 14. End with CNTL/Z.e.2.

IS-IS.0000. IS-IS. End with CNTL/Z. The other issue with maintaining secrecy is the question of “How many keys should be used in the routing domain?” That is. and EIGRP use Cisco IOS key chains. RIP. Therefore an attacker has to be physically attached to a router in the IS-IS network to maliciously disrupt the IS-IS routing environment. Version 1. However. whether one key should be used for the entire routing domain. IS-IS offers a unique security advantage compared to other IP routing protocols. A key’s secrecy is intact only if it is known by the trusted routers but hidden from any attacker. ISIS packets are encapsulated over the data link and are not carried in IP packets.1. OSPF uses single keys. are susceptible to attacks from remote IP networks through the Internet because routing protocol packets are ultimately embedded in IP packets.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.0000. an administrator must manually change the keys when their lifetimes expire.1c 115 . such as RIP.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. RIP. “Other routing protocols.” [27] Key Management The strength of these methods. Using a separate key for each router neighbor-to-neighbor connection can become an administrative nightmare. so using a common key for the entire routing domain is recommended. Key lifetime is also important. OSPF. The best method for distributing keys to trusted routers is to do it manually.255. East(config)# router isis secure-network East(config-router)# net 49.0003.1. and EIGRP routing update authentication.0001. which offer substantial control over key lifetime. which makes them susceptible to remote access by intrusive applications. or a separate key for each router neighbor-to-neighbor connection. because failure to do so can compromise the entire network. OSPF and BGP.0.20 255. depends on two factors: the secrecy of the keys and the quality of the keys.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. one per line. maintaining the secrecy of the key becomes much more important.

For more information on the internal workings of static routes.6. Central(config)# ip route 14. Central# config t Enter configuration commands. then that static route can be overridden by dynamic information.255. See Section 4.2). Also.20 120 Central(config)# end Central# 116 Version 1. if an administrative distance is specified. The following example illustrates how to create a static route with a higher administrative distance than OSPF.1. 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.2. OSPF-derived routes have a default administrative distance of 110. consult [7].5. and specify the start-time and end-time to accept and send individual keys. A dynamic routing protocol. one per line. static routes take precedence over their dynamic counterparts.1 for a detailed description. then your router security policy should define the key management procedures and responsibilities.255. The rules for generating good passwords apply to generating good keys as well. and RIPv2 authentication. Both of these are configured within the key chain configuration mode. Static routes have a default administrative distance of 1. such as OSPF. In most cases. such as link failures. static routes are very secure. These routes typically take precedence over routes chosen by dynamic routing protocols. In one sense. Static routes cannot easily handle events that change the network topology. End with CNTL/Z. can correctly re-route traffic in the case of a router or link failure. The other factor that authentication relies upon is the quality of the keys. If you use routing update authentication. The number of different keys and the key validity periods should be defined in the router security policy.0 255. Static Routes Static routes are manually configured on the router as the only path to a given destination. Using static routes exclusively can make network administration very difficult. 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. 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.1c . However. EIGRP.0 14. For example. The router must be configured to maintain correct time (see Section 4.1. but not OSPF authentication. These commands apply to keys in a keychain. so they can be used for IS-IS.

will increase network load. constantly sending routing updates will likely expose the identity of the infiltrator.Advanced Security Services The simplest approach for discarding traffic with black-hole routes is to set up static routes. It will also wait 10 seconds between two consecutive SPF calculations.1c 117 . Hello 10. 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. If this value is 0. . . Retransmit 5 Hello due in 00:00:05 . Timer intervals configured. Central# show ip ospf 1 . unless the attacker continues to send routing updates. State DROTHER. then two consecutive SPF calculations can be done without any waiting period.4. 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 . The default settings have been selected to provide optimal performance. which is 40 seconds in the example. Area 1 Transmit Delay is 1 sec.20. In either case. The output of the show ip ospf interface interface command shows that the time between Hello packets on interface ethernet0/0 is 10 seconds. Reducing both of these timers causes routing to switch to an alternate path more quickly in the event of a failure. Wait 40. different aspects of network security will be addressed. especially when using RIP. is the time hello packets must not have Version 1. The Dead interval. then reducing the convergence time will cause that false route to die quickly. The timers for OSPF can be viewed by using the show ip ospf pid command and the show ip ospf interface interface command. The example below illustrates how to reduce convergence on an OSPF and RIP network. If an attacker creates a spoofed route to redirect traffic.150/24.168. reducing convergence time. SPF schedule delay 5 secs. . Central# show ip ospf interface ethernet0/0 Ethernet0/0 is up. Priority 1 . If this value is 0. As a cautionary note. line protocol is up Internet Address 192. then Central starts an SPF calculation after receiving a topology change. as discussed in Section 4. However. Dead 40.6.

Timer intervals configured. Priority 1 . .1c . If the Hello-interval and Dead-interval are modified on a router. Central# config t Enter configuration commands. State DR. SPF schedule delay 1 secs. That is. The example below shows how to modify OSPF timers. . This time. The second modification sets the Hello-interval to 5 seconds. Central# config t Central(config)# router ospf 1 Central(config-router)# timers spf 1 4 Central(config-router)# end Central# show ip ospf . The Retransmit interval is the time between LSA (Link State Advertisement packets sent by OSPF) retransmissions. End with CNTL/Z. Hello 5. 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. Retransmit 8 Hello due in 00:00:02 .168. then all other OSPF routers on that network must be changed as well. Similarly. Area 1 Transmit Delay is 6 sec. Hold time between two SPFs 4 secs . Dead 20. Central# show ip protocols .150/24.20. The first modification sets the SPF calculation delay to 1 second and the delay between two consecutive SPF calculations to 4 seconds. the timers for RIP can be viewed by using the show ip protocols command. all routers on that network must have the same Hello-interval and Dead-interval. line protocol is up Internet Address 192. one per line. Wait 20. The Transmit Delay is the time in seconds that Central will take to transmit a link-state update packet. . must be greater than the expected round trip between Central and any other router on the same network. Otherwise. the Retransmit-interval to 8 seconds. and the Transmit-delay to 6 seconds. 118 Version 1. which is 5 seconds. the Dead-interval to 20 seconds.Router Security Configuration Guide been seen before Central declares its neighbor dead. . the routers will be sending needless LSA packets.

It is also possible to redistribute OSPF or IS-IS routes over RIP. hold down 150. is the amount of time. and this is preferable to running RIP on an entire large network. 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. which is not shown. Central# In general. . hold down 180. The hold down time is the time that a route will remain in the routing table before a new route is accepted.1. flushed after 240 . then the route is declared invalid.0.250 255. It is typically used when the wildcard specification on the network router configuration command configures more interfaces than desirable. In its current configuration. 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. an update will be delayed before transmission. Central# config t Enter configuration commands. End with CNTL/Z. The example shows how to modify the RIP timers. consult Chapter 13 of [2]. The example below illustrates such a case. Router1# show config .0 ! Version 1. OSPF and IS-IS are preferable to RIP. 4.4. .255. .Advanced Security Services .15.1.0 net ip address 14. If no update is received within 180 seconds. Central(config)# router rip Central(config-router)# timers basic 20 120 150 230 3 Central(config-router)# end Central# show ip protocols . flushed after 230 . The sleep time. interface ethernet0 description Active routing interface for 14. next due in 6 seconds Invalid after 120 seconds. next due in 22 seconds Invalid after 180 seconds.1c 119 . measured in milliseconds. It can also be used to keep any unnecessary parties from learning about the existence of certain routes or routing protocols used. Routing Protocol is "rip" Sending updates every 20 seconds. one per line.4. . Routing Protocol is "rip" Sending updates every 30 seconds. For details on this topic. RIP routing updates are sent every 30 seconds.0.

90.0.0 net ip address 14.0 ! interface ethernet1 ip address 14.1c . OSPF will run only on interfaces ethernet0 and ethernet1. RIP broadcasts will be sent through that interface. This command is especially important when using RIP version 1.0 net ip address 14.90.50 255.3.0 ! interface ethernet2 description Passive interface on the 14.150 255.0.255. interface ethernet0 ip address ! interface ethernet2 ip address 14.150 255.0.2.x. although ethernet0/0 is part of an OSPF network.0 but routing updates will still be received and processed. .2.0.x. When used on RIP. An alternative method to this is to simply not enable OSPF on certain interfaces. However. because that version only uses major network numbers.255.255 area 0 .255 area 0 network Router1# This command functions slightly differently on RIP. 120 Version area 0 passive-interface ethernet2 .Router Security Configuration Guide interface ethernet1 description Active routing interface for 14.255. i.0 0. enabling RIP on Central will cause RIP broadcasts to be sent out of interfaces ethernet0/0 and ethernet0/1. this command stops routing updates from being sent out on an interface. Thus. .15.1. this command blocks routing updates from being sent or received on an interface.0. Router1# When used on OSPF. ! router ospf 1 network 14. 14. OSPF has been enabled to run on all subnets of 14.x.3.0. The example below illustrates how to remedy that problem.0.e. by designating ethernet2 as a passive interface. The reason for this is that both interfaces appear to have the same Class A internet address.0 ! router ospf 1 network 14.0 255.250 255.0.255. In Figure 4-3. In the example above. . Router1# show config .2. as shown below.

The distribute-list out command in OSPF configuration mode stops routes from being advertised in updates.10. this restriction only applies to external routes. The following example shows how to prevent Central from advertising the 14.0 0. but it does not stop routes from being sent out in the link-state advertisements (LSAs).Advanced Security Services Central# config t Enter configuration commands. For more information about how to create access lists. Each command behaves differently with respect to the routing protocol used. however.0 network. Central# config t Enter configuration commands. since OSPF is not enabled on the interface to the RIP network.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. you must first create an access list. an access list with rules filtering out 14.10.0 network from the RIP routing domain into the OSPF routing domain. For illustration purposes.2. With this setting North and East would not see a route to the 14. use the distribute-list out command . To suppress advertisements of particular networks in updates. this step is unnecessary. End with CNTL/Z.0. However. Some authors. End with CNTL/Z. routes from a different autonomous system (AS). use the distribute-list in command.2. End with CNTL/Z. one per line. Central(config)# access-list 55 deny 14. see Section 4. The example below illustrates that. 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.10. Central# config t Enter configuration commands. 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. Version 1. This command has two primary functions.2.0. To apply this command to a routing protocol. including Parkhurst [2]. To filter the receipt of network updates.2.10. one per line. one per line.0/24 will be used.3. Therefore.1c 121 . the following example is for illustration purposes only. Thus all downstream routers will learn about the networks that were supposed to be filtered in these LSAs. advise against using distribute-list in for OSPF. that is.

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.0 network. thereby preventing North from learning that route. a similar problem arises.2.2. In either case.2. South# config t Enter configuration commands.0. End with CNTL/Z. hosts from the 14.0 network because it is directly connected to it. End with CNTL/Z. The following example shows Central deleting the route to 14. as their default gateway. However. However. network from communicating with hosts on any other subnets is to simply turn off interface Ethernet0/1 on South.2.0 network if the hosts on the latter network use South.2. South(config)# access-list 55 deny 14.10. that is.2. the easiest way to prevent hosts on the 14. as their default gateway.0 network are prevented from reaching the Internet. Therefore.0. Using the first filter. while Central is not advertising a route to the 14. End with CNTL/Z. South still maintains a route to the 14. one per line. one per line. East and Central will not see routes to the 14.2.0 0.10. This is because. The second and third filters each fix the problem that was evident with the first filter. all updates sent from that router will not advertise the deleted route. it will not advertise this network to other routers.0 network can communicate with hosts on the 14. hosts on the 14.10.10. Thus. instead of North. Central still has the route in its table.2.0 network as it comes in from a RIP update from South. instead of Central.0 network if the hosts on the latter network use Central. since Central no longer has a route to network 14.10.0 network can be made with hosts on the 14.0 network. one per line. Central# config t Enter configuration commands. North and East will not see a route to 14. 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. the effect of applying the same filter used in the previous examples to South is that North. Thus. the three different filters also have unusual side effects.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. Subsequently.) 122 Version 1. (Of course. Connections from hosts on the Security Configuration Guide Central# config t Enter configuration commands.1c . This is because either Central is filtering the routes it receives (second filter) or South filters the routes it advertises (third filter). but that is a drastic approach.

OSPF offers better convergence times. [18]. Migration procedures are beyond the scope of this document. In this particular case. [21]. consult [15]. then migrating to OSPF is the recommended solution.4. while the service provider gateway router. and is outside the scope of this guide. Exterior Gateway Routing Protocol Security Configuring an exterior gateway protocol can be very complex. [23]. This sub-section presents four security mechanisms that you should when using BGP-4: the Generalized Time-to-Live (TTL) Security Mechanism. a brief discussion of CPP appears in Section 4. our example network constitutes a single autonomous system. [17]. Both protocols are supported by virtually every routing vendor. The example below shows how to turn off RIP. and route flap damping. Control Plane Policing (CPP) can also help protect BGP operations. and using OSPF reduces the likelihood of accidentally sending out routing update packets on an unintended interface.7. However. and the service provider that serves as its connection to the Internet. The examples in this sub-section show how to configure the indicated BGP security mechanisms. Remember to turn off RIP on all the routers after migration.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. is part of AS 27701. Version 1. MD5 authentication. If support for RIP is not an essential requirement.5. check the configured protocols using the command show ip protocols. Central(config)# no router rip Central(config)# After disabling RIP. For more information on BGP and BGP Security. While both protocols support authentication. Failure to do so will not cause a routing failure. but an attacker could then take advantage of RIP and insert a malicious route into the routing table. but they do not show how to configure BGP routers in a large network. see [2] for detailed directions. prefix list filtering. but OSPF offers better scaling and faster convergence. an important step to remember is to remove RIP after OSPF has been enabled. Figure 4-6 shows the relationship between our example network. OSPF is the superior routing protocol. named ISPCust7.1c 123 . AS number 26625. and [30] through [43]. 4.3.

Otherwise.2. Generalized TTL Security Mechanism (GTSM) The Generalized TTL Security Mechanism (GTSM).e.0.0(27)S and 12.2.250 Type escape sequence to abort Tracing the route to 14. it does not address operational issues that arise when deploying BGP infrastructure in a large network. otherwise the BGP session will terminate. and [31]. 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.1c . consult [15]. Also. utilizes the Time-to-Live (TTL) field of the IP header to protect exterior BGP (eBGP) peering sessions from remote attacks.2. Since remote attacks originate multiple router hops away from an intended target.250 14. If the received IP packet contains a TTL value greater than or equal to the expected TTL value (i. documented in RFC 3682 [32] and introduced in Cisco IOS 12.Router Security Configuration Guide 14. [17].0. use the trace command to verify that the BGP peers are indeed separated by the expected number of hops. [21]. then the packet is processed. Before enabling GTSM. 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. The following example shows how to implement the GTSM between routers North and ISPCust7 in Figure 4-6.250 124 Version 1. the packet is silently discarded. Additional information on GTSM may be found in [33].0.3(7)T. 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. North# trace ip 14. This mechanism uses the TTL value in a received packet and compares it to an administrator defined hop count.2. For more information on BGP operations.0.

2. This security mechanism operates between two BGP peers and requires the configuration of a shared key on each of these peers. North (config)# router bgp 26625 North(config-router)# neighbor 14.2.2. BGP shared keys should follow the password guidance specified in section 4. “Protection of BGP Sessions via the TCP MD5 Signature Option” [34].20 Type escape sequence to abort Tracing the route to 14.2.e. one-way hash) for each segment transmitted and received during the BGP session respectively.0.250 ttl-security hops 1 North(config-router)# end North# ISPCust7# trace ip 14. The shared key takes the form of a password configured on each peer router.Advanced Security Services 1 14. one per line.2.0.20 1 14.0. North(config)# router bgp 26625 North(config-router)# neighbor 14. ISPCust7# config t Enter configuration commands. and man-in-the-middle attacks by implementing RFC 2385.2. End with CNTL/Z. one per line. End with CNTL/Z.1c 125 .0.250 0 msec * 0 msec North# config t Enter configuration commands. End with CNTL/Z. 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. ISPCust7(config)# router bgp 27701 Version 1. one per line.1.250 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. North# config t Enter configuration commands. 0 msec * 0 msec ISPCust7# config t Enter configuration commands. End with CNTL/Z. ISPCust7(config)# router bgp 27701 ISPCust7(config-router)# neighbor 14. The shared key is used to create and verify the MD5 signature (i.20 ttl-security hops 1 ISPCust7(config-router)# end ISPCust7# BGP and MD5 Authentication BGP peers can be protected from DoS.0.0.5 and in the relevant IOS documentation The following two-part example pertains to Figure 4-6.250 remote-as 27701 North(config-router)# neighbor 14. The commands below would be performed on North by the local administrator. spoofing. one per line.0.

3.) Prefix list filters can be used to filter both inbound and outbound route updates on a per-peer basis. and prefix hijacking caused by malicious prefix advertisements.e. prefix list filtering is an important technique employed by AS network administrators to filter BGP route advertisements. and others maintain bogon address lists. AS ingress and egress prefix lists should filter martian address space (i. 4. address space that should never be propagated through the Internet). ISP ingress prefix lists applied to advertisements from a customer should permit only those prefixes that have been assigned or allocated to the customer. 2. Although the use of prefix list filters is not restricted to BGP. There are several general rules that you should consider when developing prefix lists and filters for your BGP routers: 1. Cisco.20 remote-as 26625 ISPCust7(config-router)# neighbor 14. 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 prevent ISP core network prefixes from being advertised to any other AS. address space that IANA has not yet allocated). At a minimum. 5.e.2.20 password r0utes4All ISPCust7(config-router)# end ISPCust7# Prefix List Filtering Prefix list filtering is a common technique used to prevent damage such as DoS.0. Team Cymru. ISP ingress prefix lists should deny advertised prefixes more specific than /24.1c .Router Security Configuration Guide ISPCust7(config-router)# neighbor 14. Use prefix list filtering to enforce routing policy by disregarding advertisements and withdrawals of specific prefixes. 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]). ISP egress prefix lists should filter what is advertised to other ISP peers. 126 Version 1. 6. AS ingress prefix lists should prevent internal prefixes from being advertised to the AS by some other AS. examples appear in [18] and [42].0. (Prefix filtering may be used to prevent an ISP from transiting traffic through a multihomed customer to another ISP. traffic redirection.2. including both RFC 1918 special use addresses [44] and bogon addresses [38] (i.

0.0. North# config t Enter configuration commands. It shows how to configure and apply prefix list filters to prevent a subset of martian addresses.Advanced Security Services The following example pertains to Figure 4-6. refer to [35] and [42]. Version 1. Damping controls the effect of route flapping which occurs when a route constantly transitions from an up-to-down or down-to-up state.0. from being advertised by router North and accepted by router ISPCust7.250 prefix-list NO-MARTIANS out North(config-router)# end North# ISPCust7# config t Enter configuration commands. 10.16.0. notably private addresses (i.0/0 le 32 ISPCust7(config)# router bgp 27701 ISPCust7(config-router)# neighbor ISPCust7(config)# ip prefix-list NO-MARTIANS seq 70 deny For more information on prefix filtering.168.0. End with CNTL/Z.168.0.20 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. The explicit permit all prefix list entry with sequence number 100000 is used so that after the deny filters are applied.0/8 North(config)# ip prefix-list NO-MARTIANS seq 70 deny 172.0/16 ISPCust7(config)# ip prefix-list NO-MARTIANS seq 100000 permit 0.0. These transitions cause excessive BGP route update messages (i.0. one per line.0/12.1c 127 . ISPCust7(config)# ip prefix-list NO-MARTIANS seq 60 deny 10.0. one per line. and 192.0/12 North(config)# ip prefix-list NO-MARTIANS seq 80 deny 192.0/16 North(config)# ip prefix-list NO-MARTIANS seq 100000 permit 0.0. North(config)# ip prefix-list NO-MARTIANS seq 60 deny 10.e.0.0/0 le 32 North(config)# router bgp 26625 North(config-router)# neighbor 14. End with CNTL/Z. 172.0.0/8.0/16). no other prefixes will be affected by the implicit deny of the prefix list filter mechanism.2. ISPs and other backbone providers may configure BGP damping to mitigate route flapping. add/withdraw routes) to propagate through the network.16. 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.0/12 ISPCust7(config)# ip prefix-list NO-MARTIANS seq 80 deny 192.e.

The fine granuality access lists provide can impose significant administrative and performance burdens.2 and earlier ! 12. Access lists provide the administrator with a high degree of precision in selectively permitting and denying traffic.range is 1-20000. North# show ip bgp dampened-paths ! 12. Cisco default parameters are shown below. For example. Using Black-Hole Routing Many administrators configure their routers to filter connections and drop packets using basic and extended access lists. Consult the following references before implementing route flap damping: [23]. and monitor your network carefully to determine the appropriate parameters for your network environment.3 or North# show ip bgp dampening dampened-paths 4.Router Security Configuration Guide The syntax for the bgp dampening command permits several optional parameters. bgp dampening [half-life] [reuse] [suppress-limit] [maxsuppress-time] • half-life . Opinions differ on the usefulness of route flap damping and the correct values for damping parameters. The list of dampened paths is useful in determining which remote networks are having instability problems. default is 15 minutes.4. depending on the network 128 Version 1. • reuse . If you choose to use route flap damping. [37].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.range is 1-225 minutes. To display the dampened routes with the corresponding suppression time remaining.range is 1-45 minutes.g.1c . The following example shows the network administrator of ISPCust7 in AS 27701 enabling route flap damping with default parameter values. [36]. • suppress-limit . • max-suppress-time . access lists would allow an administrator to block only Telnet (TCP port 23) traffic from exiting their network.6. use one of the two commands shown below.2.range is 1-20000. 60 minutes). default is 750. Operational experience has shown that vendor defaults may be too aggressive. begin by using the defaults. ISPCust7(config)# router ISPCust7(config-router)# ISPCust7(config-router)# ISPCust7(config-router)# ISPCust7# bgp 27701 neighbor 14.0. default is 2000. default is 4 times the value of the half-life parameter (e.

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

How Does Unicast Reverse-Path Verification Work? All routers maintain a routing table that lets them decide how to forward packets.1. Because the packet has arrived on the wrong interface.Router Security Configuration Guide null routes throughout a network.2. Null routing can also be combined with filtering to support traceback of some types of DoS attacks.10.C. which is not the interface on which packet 2 has arrived. and in situations where it applies. it is from a host behind the South router. However.B. the router uses its routing tables to decide whether to accept or drop individual packets arriving on the interface. 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. 14. for a detailed look at the topic.3. Additional practices for null routing are described in [15]. Setting up such a capability is beyond the scope of this guide. the configuration of such advanced null routing is beyond the scope of this guide.1c . It is also possible to set up automatically triggered null routing in an entire network. and in some cases it can offer significant advantages over using access lists for that purpose. the router looks up its source address. in the routing table.12. otherwise it is ‘bad’.7. it is good security practice to reject a packet with a spoofed source address. Figure 4-7 shows two packets arriving at the router Central on its ‘inside’ interface. As noted in Section 4.4. the router looks up its source address. It finds the interface Eth 0/1.20. Unicast reverse-path verification supports rejecting such packets. Eth0/1.D was received is the one that the router would use to send a packet to A. For packet 1.B. It finds interface Eth 0/0. so the router forwards packet 1 normally out interface Eth 0/0. 130 Version 1. Unicast reversepath verification is not enabled by default. When this feature is enabled on an interface. consult [45].C. This is a match. The first packet bears a proper source address.2. For packet 2. it fails unicast reverse-path verification. as described in [24]. Good packets are forwarded normally. it might have been generated by a piece of malicious software secretly installed somewhere on LAN 2. in the route table.0 and later support a routing-based filtering feature called IP unicast reverse-path forward (Unicast RPF) verification. Used correctly. and the router discards the packet. 4. you must explicitly apply it to each interface where you want verification to be done. which is the interface on which packet 1 has arrived. 7. then the packet is considered ‘good’.D. bad packets are discarded. The second packet bears a spoofed source address. unicast RPF verification prevents most forms of IP address spoofing. Unicast Reverse-Path Forwarding Verification Most Cisco routers running IOS 12.

1.20 -14.9. then simple unicast RPF verification may not be used.0/16 0.12. Access lists.9. If your router does not or cannot support CEF. dest=7.10. Avoid unicast RPF verification if any of the following conditions apply.64 14.0/24 Pa c ke t 2 Interface Eth0/0 Trash Interface Eth0/1 Packet 1 src=14.1c 131 .Advanced Security Services 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.0. but only when the network architecture permits it to be used.1. • Router uses asymmetric routes – if any of the interfaces on the router participate in asymmetric routes (one interface for sending. it automatically adjusts to most changes in network structure. 14.0/24 14. while more broadly applicable. Eth0/0 14. • Router does not support CEF – according to the Cisco documentation.0.9 Packet 2 src=7. Version 1. then you cannot use unicast RPF.1. use access lists instead.0/24 14. also require more maintenance.0.2 dest=10.0/24 14. When to Avoid Unicast Reverse-Path Verification This facility can be very useful for rejecting packets with improper IP source addresses.2.1. and a different one for receiving). unicast reverse-path verification depends on Cisco Express Forwarding.20 Packet 1 Destination Gateway Interface 14.0/0 -14.2.250 Central Eth0/1 14. It will incorrectly reject packets arriving on the receive leg of the asymmetric route.9. Cisco has stated that future IOS versions will perform unicast RPF correctly in these cases [11].5.

132 Version 1. and then enable verification on the desired interfaces.3 and later). then the access lists are applied. a filtering router between a LAN and the Internet). Note that you must not turn off CEF while unicast RPF is enabled.255. Under 12.15. Unicast Reverse-Path Verification and Access Lists Cisco IOS version 12. Thus. the access list allows you to create exceptions to unicast RPF’s usual functioning.1. If the access list permits the packet. Configuring Unicast Reverse-Path Verification Unicast RPF verification depends on a particular routing mode called Cisco Express Forwarding (CEF). In particular. only the lines about drops may appear. Central# show ip interface eth0/0 Ethernet0/0 is up. access lists may be applied to RPF. IP verify source reachable-via RX. or to view statistics about dropped packets. enter interface configuration mode and use the command no ip verify unicast reverse-path (12. allow default 0 verification drops 0 suppressed verification drops Central# To disable unicast RPF.2 and earlier.255 . 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.1 and later include significant enhancements to unicast RPF. to use unicast RPF. 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.3. first enable CEF. Consult [10] for details about CEF requirements. The example below shows how the output will look when unicast RPF is enabled in IOS 12.1c . If the access list denies the packet. line protocol is up Internet address is 14. To check whether unicast RPF is enabled on a particular interface.255. Used properly. Also. use show ip interface interface-name. then it is dropped.g. then it is forwarded. consult [10]. .2) or no ip verify unicast (12. When a packet fails reverse-path verification. For more details on how and where to apply unicast RPF verification.0 through 12.250/16 Broadcast address is 255.Router Security Configuration Guide Unicast RPF verification is best suited for routers that act as part of the security boundary between two networks (e. Therefore.

OSPF Network Design Solutions. [5] Thomas. OSPF – Anatomy of an Internet Routing Protocol. This book offers a good overview of IP routing and related topics. Prentice Hall. McGraw-Hill. [8] Rudenko. J. W.. Version 1. For more information about advanced unicast RPF features. consult [11] and [15]. 1998.. Laura. The most comprehensive and readable guide to the TCP/IP protocol suite.R. W.8. 1999. McGraw-Hill. T. 1999. security.T. [2] Parkhurst. Cisco Routers for IP Routing: Little Black Book. Coriolis Group.Advanced Security Services if the access list rule that denies a packet includes the log qualifier (see Section 4. with lots of practical advice. Includes discussion of design issues. An excellent introduction to basic Cisco IOS tasks. too. including routing and routing protocols. Editor.. References [1] Albritton. A very practical and pragmatic guide to setting up routing protocols. implementation. [6] Stevens. IP Routing Protocols.1) then a log message is generated. J. 1998.1c 133 .. 4. I. Cisco Router OSPF . U. Includes a good section on troubleshooting. Cisco Press. Advanced Cisco Router Configuration. 1999. [4] Moy. A very good survey of routing protocols and the technologies behind them. Addison-Wesley. [3] Black. TCP/IP Illustrated. Cisco IOS Essentials. 2000... great technical background for any network analyst.3.Design and Implementation Guide. 1998.4. Volume 1. Addison-Wesley.R. [7] Chappell. Portions of this book that are particularly relevant to Routing Protocols are Chapters 2 and 7. Detailed analysis of OSPF and MOSPF. Comprehensive and practical guide to OSPF use. Cisco Press.M. 1994. A great reference book for a variety of Cisco configuration topics. and also explains how to configure Cisco routers for OSPF in a variety of situations. and deployment..

[13] Smoot. Cisco Internetworking Technology Overview.pdf Basic IOS documentation on unicast reverse-path forwarding verification. available at: http://www. “RIP and OSPF Redistribution”. 2000. ito_doc/routing. Cisco Systems. The IETF specification for ARP. Cisco Press. Cisco IOS 12.. 2000. Cisco Internetworking Case software/ios120/12cgcr/secur_c/scprt6/scrpf.1(2)T Release Notes. this document gives a broad. all with fully worked-out examples. April 2002.1c . P.0 Security Configuration Guide. available at: http://www. RFC 826. 2000. Cisco ISP Essentials. performance-oriented view of IP routing. and Smith.1 releases. Cisco Systems. [14] and Quarterman. Cisco Systems. M&T Books. This detailed Cisco guide for Internet Service Providers includes extensive discussion of routing protocols (especially BGP). B. RFC 1027. Cisco Systems. [12] [10] “Configuring Unicast Reverse Path Forwarding”.Router Security Configuration Guide [9] Cisco Systems. Cisco Router Troubleshooting Handbook. includes a good explanation of the concepts. available under software/ios121/121newft/121t/121t2/rpf_plus. 1982. [15] Greene. available at: http://www. D. 1st Edition. P. J.. The IETF specification for Proxy Documentation for new Unicast RPF features that are being integrated into IOS 12.. 2002. C-M. 134 Version 1. [11] “Unicast Reverse Path Forwarding Enhancements”. This pragmatic volume offers good advice for diagnosing and correcting problems with routing and routing protocols. Cisco IOS 12. and an in-depth treatment of Unicast “An Ethernet Address Resolution Protocol of Converting Network Protocol Addresses to 48-bit Ethernet Address for Transmission on Ethernet Hardware”. [16] “Routing Basics”. “Using ARP to Implement Transparent Subnet Gateways”.htm As a general overview of routing concepts and terminology. 1987.

[19] Doyle.. J. RIPE-229.Advanced Security Services [17] Stewart. “BlackHole Route Server and Tracking Traffic on an IP Network”.Inter-Domain Routing in the Internet. an excellent reference for network engineers responsible for enterprise design. Cisco Systems. Addison-Wesley. The bookemphasizes techniques for designing efficient networks. Schmitz. available at: http://www. October 2005. 1998. Version 1. Cisco Press.0. Provides a basic understanding of routers and routing protocols through a thorough inspection of IP interior gateway routing protocols.. [22] Cisco IOS IP Configuration Guide. Cisco BGP-4 Command and Configuration Handbook.html This note describes rationale for BGP route flap dampening. J. and access control lists to trace back the sources of some DoS attacks. W. Version 2. 1999. [18] Thomas. Cisco Press.2. RIPE. 2002. This volume of the Cisco IOS documentation offers extensive information on configuring all of the routing protocols discussed in this section. [21] Parkhurst. and Vistoli. available at: http://www. [20] “Enhanced IGRP”. [23] Panigl.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. available at: http://www. A white paper that describes the features and operation of EIGRP. “Secure BGP Template Version 4. BGP4 . IOS 12. [24] Morrow. “RIPE Routing Working Group Recommendations for Coordinated Route-flap Damping Parameters”.W. A clear.ripe. Cisco Press. 2002.. 2001. Smith. October 2001.1c 135 . The command guide provides very good configuration. concise resource for Cisco IOS software BGP-4 commands.secsup.Volume 1. C.1”.R. Cisco Internetworking Technology Overview.. R. and includes links to examples for Cisco This terse and technical note describes a technique for using null routing. Routing TCP/IP . troubleshooting and verification guidance. Provides a good overview of BGP and practical advice on using it.

“Protection of BGP Sessions via the TCP MD5 Signature Option. Cisco Press. Cisco Technologies Documentation.3(7)T. Heasley. [33] “BGP Support for TTL Security Check”. A comprehensive hands-on guide for resolving IP routing problems.Second Edition. Troubleshooting IP Routing Protocols. 136 Version 1. 2003. A.. and Bartell. Good reference describing the IOS command set for HMAC-MD5 authentication. Cisco Systems.1c . F. 2002. and Meyer..html This page offers access to a large number of technical documents and examples about IOS-supported routing protocols. 2003. IS-IS Network Design Solutions. 2004. including guidance on multiprotocol extensions. and 12.” RFC 2385. August 1998. [29] A discussion of advanced BGP topics applicable to both enterprises and ISPs. 2002. BGP Design and Implementation. 2001. A reference for IOS configuration of the GTSM. M.0(22)S.2(25)S.. [34] Heffernan. [30] Halabi. The definitive IS-IS reference and design guide. available under: http://www. [28] “Configuring IS-IS Authentication”. A. Cisco IOS Release Notes for 12. available at: 2001. organized by protocol. A thorough presentation of inter-domain routing concepts.” RFC 3682. and scenarios.. Cisco tsd_technology_support_protocol_home.0(27)S. [26] “IS-IS HMAC-MD5 Authentication and Enhanced Clear Text Authentication” IOS Release Notes for 12. Cisco Systems. The IETF specification for GTSM. [27] Martey. S. [31] Zhang. Cisco Press.0(21)ST and 12. “The Generalized TTL Security Mechanism (GTSM). Cisco Security Configuration Guide [25] “IP Routing Protocol Groups”. R. Internet Routing Architectures .pdf Provides a good overview of IS-IS authentication with an emphasis on plaintext authentication. 12. issues. Cisco Press. Cisco Press. February 2004. [32] Gill. Cisco Technology White Paper.. containing command syntax and a detailed configuration procedure.

Also includes some explanatory notes. “Flap Damping .org/mtg-0210/ppt/flap. [39] “ARIN 2005 Statistics”. [38] “The Team Cymru Bogon Reference Page”.html A list of minimum practical allocations by APNIC. Team Cymru.” Proceedings of the RIPE50 Conference.html A list of minimum practical allocations by RIPE. with links to detailed lists and historical data. [37] Smith.ripe. presentations/ripe50-plenary-wed-flap-damping.nanog. Cisco course notes.pdf A good overview of prefix filtering with examples and references. 2005.pdf A conference presentation that questions the value of route flap damping.Advanced Security Services The IETF specification for MD5 protection of BGP sessions. ISP Security Boot Camp.html#cidr An overview of address space allocated by ARIN. [36] Bush. L. available at: http://www. 2005. for use in prefix list filtering.. available at: 2005. B. and Version 1. RIPE 349. 2002.arin. RIPE. Cisco Systems. available at: Ingress-Prefix-Filter-Templates/Prefix-List-Filtering-Mar-03. ripe-ncc-managed-address-space.Where to now?.com/Bogons/ An excellent resource about bogons and bogon filtering. available at: ftp://ftp-eng. [40] Vegoda.1c 137 . This page also provides a well-maintained list of bogon address ranges in a variety of formats. available at: http://www. Asia Pacific Network Information Center.pdf A presentation of flap damping issues and a call for a response to deal with those issues. Griffin..ripe. P. “Address Space Managed by the RIPE NCC”. available at: http://www. [35] Greene. available at: http://www. “Route Flap Damping: Harmful?” NANOG presentation. for use in prefix list filtering. 2002. [41] “Allocation sizes within APNIC Address Ranges”.

[45] “Remotely Triggered Black Hole Filtering – Destination Based and Source Based”. B. 138 Version 1. [44] “SAFE: Best Practices for Securing Routing Protocols”. version 1. Cisco Systems. “Ingress Prefix Filter Templates”.pdf A recent and detailed paper about threats to routing protocols and means to mitigate those threats. [43] prodlit/sfblp_wp. to be used for filtering martians. February 1996..Router Security Configuration Guide [42] Greene. docs/blackhole. “Address Allocation for Private Internets”. RFC 2005. Maintained as part of the bogon project [38]. This RFC describes the reserved IP address allocations for private intranets. B. Cisco Ingress-Prefix-Filter-Templates/ IOS command templates for ingress prefix filter lists. Cisco white paper. 2004. Moskowitz.4. Cisco Systems. available at: http://www. available under: available at: http://www. and Rekhter. July 2005.pdf An in-depth description of how to set up centrally-triggered null routing using your internal BGP mesh.1c . Cisco white paper.

and must also be disabled or carefully configured. and update facilities offered in Cisco IOS 11. Audit and Management 4.0. because new releases include fixes for security vulnerabilities. Cisco routers and Cisco IOS are designed to support centralized audit and management. • Software Maintenance – Keeping up with new major software releases is important. Cisco routers have the ability to log a great deal of their status. especially the Cisco IOS Configuration Fundamentals Command Reference documents for IOS 12. management.5. Installing new Cisco IOS software in a router is not especially difficult.5. describes the logging configuration commands. This section describes the logging. Concepts and Mechanisms Routers are a critical part of network operations and network security. The sub-sections below describe recommended configurations for good security. troubleshoot service disruptions.2. and react to probes and scans of your network. • Logging – Cisco routers support both on-board and remote logs.1c 139 . 12.Advanced Security Services 4. • Network Monitoring – Cisco routers support basic facilities for Remote Network Monitoring (RMON).5. and aid in the analysis of suspected security breaches. and presents some configuration examples. good logs can help you find configuration errors.0. Configuring Logging and Time Services Logging is a critical part of router security. 4.3.1. improve security. and later. • Debugging and Diagnostics – Troubleshooting router problems requires proficiency with Cisco’s diagnostic commands and debugging features. Version 1. • Time – Accurate time is important for good audit and management. Careful management and diligent audit of router operations can reduce network downtime. The RMON features depend on SNMP. Cisco routers support the standard time synchronization protocol. Complete details on the commands and features discussed may be found in the Cisco IOS documentation. monitoring. NTP. understand past intrusions. this section explains the different logging facilities. SNMP must be disabled or carefully configured for good security. • Network Management – The standard protocol for distributed management of network components is the Simple Network Management Protocol (SNMP).

can be configured to receive log messages. access list matches. Console logging – Log messages are sent to the console line (see Section 4. The buffered data is available only from a router exec or enabled exec session. Four such events are: changing EXEC privilege level. and saving a new configuration to the NVRAM. changes in network and interface status. Log messages can be directed in five different ways. Messages can be sent to all five. Overview and Motivations for Logging Cisco routers can log system errors. 3. Cisco routers fully support the standard Network Time Protocol (NTP). and stores them in files or prints them according to a simple configuration file. The most valuable forms of logging are forms that are persistent. Console logging is handy for operators when they use the console. A syslog service simply accepts messages.3) • Recording changes in interface and network status • Recording router cryptographic security violations (see Section 5.1. 1.Router Security Configuration Guide Keeping the correct time on a router is also important for accurate logs. Terminal Line logging – Any enabled exec session. and many more kinds of events. This form of logging is useful. 4. messages printed to the console are not stored by the router. or any combination. but does not offer enough long-term protection for the logs. Buffered logging – Cisco routers can store log messages in a memory buffer. that can be preserved over time. This form of logging is not persistent.1c . • Recording router configuration changes and reboots • Recording receipt of traffic that violates access lists (see Section 4. but are otherwise of little value unless some other device or piece of software preserves the output. Some motivations for keeping router logs are listed below. 2. Turning on line logging is useful only for the operator using that line. This form of 140 Version 1.2).2) There are some events that can be important to security but which Cisco routers cannot log. on any line. Configuration guidance for NTP appears at the end of this sub-section. and it is cleared when the router boots. Syslog logging – Cisco routers can send their log messages to a Unix-style syslog logging service. which is used on the Internet and on all major US DOD networks to distribute accurate time. changing a password. This form of logging is not persistent. as discussed below. changing the configuration via SNMP. login failures.

you must use the level name. Cisco routers can generate Simple Network Management Protocol (SNMP) trap messages. It has a severity level of 5. because it can provide protected long-term storage for logs. 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. SNMP trap logging – For some kinds of events.Advanced Security Services logging is the best available for Cisco routers. up to down. in IOS 12. as shown by the numeric field “-5-” in the message name. This facility allows routers to be monitored as part of an overall SNMP-based network management infrastructure.6) Message text Message name and severity level Time that message was generated Figure 4-8: Format of a Cisco IOS Log Message Version 1.0 you may use the name or the number.1c 141 . the more critical the message is. The lower the severity level number.3 and earlier. The severity levels are described in the table below. the message below appears in the log when a user changes the running configuration. when you are using logging levels in commands in IOS 11.2. Cisco IOS messages are categorized by severity level. 5. Note that. or down to up Packet denied by an access list on an interface Appears only when debugging is enabled For example.9. Mar 31 9:00:16 EST: %SYS-5-CONFIG_I: Configured from console by vty0 (14.

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

Central(config)# interface eth 0/1 Central(config-if)# ! shutdown will log a message.1c 143 .3. . declare it to be a monitor.9.2. End with CNTL/Z. one per line. set the severity level for terminal line monitor log messages.6 Central# config t Enter configuration commands. 0 messages logged Buffer logging: level informational. 332 message lines logged Logging to 14.9. First. level 5 Central(config-if)# shutdown Central(config-if)# Mar 28 15:55:29 EST: %LINK-5-CHANGED: Interface Ethernet0/1. while using a particular line. 302 message lines logged Log Buffer (16000 bytes): Mar 28 11:31:22 EST: %SYS-5-CONFIG_I: Configured from console by vty0 (14.0 overruns) Console logging: level critical. For more information consult the tools list in Section 9. 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. Central# show users Line User Host(s) Idle Location *130 vty 0 bob idle 00:00:00 14.9. and several free and commercial syslog servers are available for Windows NT/2000/XP. this needs to be done once per session.1 flushes. There are two parts to setting up terminal monitor logging. this needs to be done only once.6) .Advanced Security Services Syslog logging: enabled (0 messages dropped. The example below shows how to set up terminal line monitoring for informational severity (level 6) on a telnet session virtual terminal line. Version 1.2. All Unix and Linux operating system configurations include syslog servers.2. one per line.6. Second. 1 messages logged Trap logging: level debugging. changed state to administratively down Setting up Syslog Logging Syslog logging is the most useful form of logging offered by Cisco routers. 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. End with CNTL/Z.

but typically only one or two are needed (see below).2. A syslog client is a host that generates messages.250 Eth 0/1 LAN 2 14. Syslog servers also support the notion of severity levels.9. the levels have the same meanings as the Cisco severity levels listed in Table 4-3 above.15. The source interface is the network connection from which the syslog messages will be sent. 144 Version 1.250 14.6 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.1. LAN 1 Eth 0/0 14. Set the severity level limit for messages sent to syslog using the logging trap command.Router Security Configuration Guide Review of Syslog Concepts A syslog server is a network host that accepts messages and processes them. The destination host may be specified with host name. use the loopback interface if you have defined one. and the source interface for the messages. the log severity level.10.2. The diagram below shows a typical configuration with syslog in use. otherwise use the network interface closest to the syslog server.2. There are several dozen valid syslog facility names.9. The severity level for syslog messages is usually the same as that for buffered log messages. The syslog facility is simply the name you’ll use to configure storage of your messages on the syslog server.64 Central log messages log messages 14. but the ones used for routers are typically local0 through local7.64 Eth 0/0 14. or an IP address.2. a DNS name. the syslog facility .9. Any number of syslog hosts may be specified. for more information consult any Unix syslog.conf(4) manual page or other syslog documentation on the server host.1c .

log Additional Issues for Syslog Logging For a router whose security is critical.conf. separate the devices into 2-5 populations with similar duties. set up access control lists to reject syslog traffic from the outside network. to load informational severity and above (level 6) messages to the syslog server.2.10. The example below shows the syslog configuration line for saving Central’s messages into a file. 31 messages logged Logging to 14.3. 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.2. End with CNTL/Z. Save all messages of critical (level 2) severity and above to a single special file. Central# Central# config t Enter configuration commands. Configuration file syntax for syslog servers is uniform for all Unix and Linux syslog servers.0 0. 0 overruns) Console logging: level notifications.log local6. Version 1. one per line.255 eq syslog In a situation where a sizable set of routers and other devices are sending messages to the same syslog server.0.Advanced Security Services The example below shows how to configure the router Central.255 eq syslog access-list 120 deny udp any 14. using syslog facility local6 and the loopback interface. it is best to designate two independent syslog servers. and local4 for LAN switches and other network hardware. local6 for border routers. On a border router. local5 for interior routers.debug /var/log/routers. rather than denying specific ports as shown here). For example.0. The syslog configuration lines below illustrate this. .0. # Save router messages to routers.6.1c 145 . Syslog uses UDP port 514.0.0 0. 11 flushes. 35 messages logged Buffer logging: level informational.6 Central(config)# logging facility local6 Central(config)# logging source-interface loopback0 Central(config)# exit Central# show logging Syslog logging: enabled (0 messages dropped. For more information on access lists. Central(config)# logging trap information Central(config)# logging 14. and otherwise save messages for each facility into a separate file. At least one of the two syslog servers’ logs should be backed up to permanent storage (CD-R or tape). Central# It is important to configure the syslog server to store router messages in their own file. the configuration file is almost always /etc/syslog. 35 messages logged Monitor logging: level debugging. consult Section 4. access-list 120 deny udp any 14. Use a separate syslog facility name for each population.2. such as a border router on the Internet. 28 message lines logged . shown in the figure above.

trap messages about link status changes and other typical network hardware events should be interpretable by commercial SNMP tools.log local4.crit /var/log/net-critical.debug /var/log/border-routers. and then keep the time synchronized while the router is in operational use. set the SNMP source interface.1.log local6. SNMP is described in more detail in the next sub-section. All Cisco routers have a clock that maintains the time and date.debug /var/log/other-net-hw. adjusting the time on each router and host on a network manually on a regular basis. It may be necessary to add Ciscospecific format specifications to the SNMP tools.crit /var/log/net-critical.log # All other router and switch messages to their respective files local6.crit /var/log/net-critical. There are four parts to setting up SNMP trap logging. and no router can keep accurate time by itself over weeks and months of operation.log local5.1c . First. it may be useful in a network that already has SNMP management deployed. although some older Cisco models may lose time when turned off. The example below shows how to configure SNMP trap logging for a receiving host 14.log local4.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.9. and may be useful in monitoring the network status. It is possible to perform manual network time synchronization. While only a small subset of all log messages can be reported this way. Time Services. and unreliable. third. error prone. Central(config)# logging trap information Central(config)# snmp-server host 14.2. select an SNMP logging host.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. Central# config t Enter configuration commands. 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. Cisco routers fully support automated network time synchronization based on the standard Network Time 146 Version 1. last. second.debug /var/log/inner-routers.2. enable SNMP traps for syslog logging. one per line. Manual time synchronization is tedious.log SNMP Trap Logging Cisco routers have the ability to report certain events as SNMP traps.9. However. set the trap logging level.log local5. End with CNTL/Z.

and so on. like an atomic clock or GPS. second. Typical NTP deployment is hierarchical. especially for intrusion and forensic analysis. NTP should be used to synchronize all the devices and hosts on a network whenever it is available. Central(config)# clock timezone GMT 0 Central(config)# exit Central# Review of NTP Concepts NTP is the standard Internet protocol for time synchronization. Examine the clock using the show clock detail command. check the clock. US east and west coasts) then you should set the router time zone on all your routers to universal time or GMT. and it is used on most large operational networks. set the time in privileged EXEC mode by using the clock set command. then set the time zone using the clock timezone configuration command. one per line. If the detail output reports a time source of NTP. Stratum 2 hosts get their time from stratum 1 servers. If the timezone is not correct. End with CNTL/Z.1c 147 . and explain how to configure it on Cisco IOS. The sub-sections below give some background information on NTP. 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 17:27:34.747 UTC Tue Mar 28 2000 Time source is user configuration Central# config t Enter configuration commands.Advanced Security Services Protocol (NTP). follow these three steps: first. see the descriptions of NTP below.g. then do not set the clock manually. Central# show clock detail 22:26:21. and last set the time.495 EST Tue Mar 28 2000 Central# If you manage routers spread across several time zones (e. as shown in Figure 4-10: one or more stratum 1 servers get their time from an authoritative source. Otherwise. NTP is designed to make time synchronization automatic and efficient. and turn off NTP on each interface using ntp disable. Setting the Time Manually To set the time. Version 1. End with CNTL/Z. one per line. Central# config t Enter configuration commands. Because having accurate time can be important for security. set the timezone if necessary.

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

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.2.9 9 11 512 377 2.2.11. The example below shows how to set up an NTP server.250 source loopback0 South(config)# exit South# ! wait one minute or so.2. Version 1. set the NTP source interface.Advanced Security Services Configuring Basic NTP Service To set up a Cisco router to participate in an NTP network. South# config t Enter configuration commands. simply designate one or more NTP servers. South# show ntp associations address ref clock st when poll reach delay offset *~14.3.candidate. designate one or more NTP servers. or the one that is the fewest number of network hops distant from the servers. second. and restrict NTP transactions to that server alone. # master (unsynced).. check the Cisco IOS documentation for details.9. clients. A full explanation of NTP access control is outside the scope of this guide.2.141 eq ntp NTP access lists can be used to impose fine-grained access control on NTP servers.0 -0. The example below shows how to configure the router South to use the router Central as its NTP server.250.141 eq ntp access-list 120 permit udp host 14. The example below shows access list entries that permit NTP traffic between router South’s loopback0 interface and a designated address of 14. and peers. consult the network administrator.2. If you use NTP. Use the source qualifier to bind the NTP service to the loopback interface. South(config)# interface eth0/0 South(config-if)# no ntp disable South(config-if)# exit South(config)# ntp server 14. To find out the main NTP servers on the wide-area network you plan to join.9. use the network interface on the same network as the designated server. one per line.11. End with CNTL/Z. For more information about access lists consult Section 4. ~configured South# show clock detail 09:30:08. The NTP source interface is the network connection from which the NTP control messages will be sent. access-list 120 permit tcp 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.250 eq ntp host 14.2. 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.250 eq ntp host 14.9. .. * master (synced).1c 149 . + selected.203.250 26. and how to check that the NTP association is working.

2.6.25 * master (synced). To restrict the set of hosts to which your router will provide NTP service. one per line.9. one per line.250 26. # master (unsynced).9. syslog logging. and authenticated network time synchronization. Unless your network is responsible for providing time service to other networks.9.1c . South(config)# ntp server 14. The host receiving the log messages is 14. Setting up such a community is outside the scope of this guide.9.250 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. ~configured South# By default. as shown here. The example below shows how to disable NTP server facilities on an interface. Central(config)# interface eth 0/2 Central(config-if)# ntp disable Central(config-if)# end Central# End with CNTL/Z. use an access list. + selected.250. Central# config t Enter configuration commands.2. End with CNTL/Z. and the time server is 14.2.9 9 11 512 377 2. This sample is formatted as it would appear in a configuration text file stored on a host for download to the router South.2.0 -0. Configuring NTP Authentication Cisco IOS supports authenticated NTP.2. one per line.9.Router Security Configuration Guide South# config t Enter configuration commands. 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. End with CNTL/Z.candidate.15. 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. you should disable NTP on all external interfaces. . which uses pre-placed keys to establish a trusted community of NTP servers and peers. buffered logging. South# config t Enter configuration commands.203.9. South(config)# ntp authenticate South(config)# ntp authentication-key 1 md5 router South(config)# ntp trusted-key 1 South(config)# ntp server 14. does not prevent a router from responding to NTP queries from other network hosts. 150 Version 1.2.250 key 1 source loopback0 South(config)# exit Note that configuring NTP authentication.250 source loopback0 South(config)# access-list 21 permit host 14.

or object.9. dotted.250 key 1 source loopback0 4. An OID is a unique. the SNMPv3 agent used in the creation of this section). SNMP may also be used by an entity on the network to send alerts indicating Version 1. A software application on a PC or workstation normally acts as the manager. one may use the OID or the actual name of each branch (separated by dots).5.6 logging on ! a tiny access list to permit access only for Central access-list 21 permit 14. If the referenced value is not a bottom leaf of the tree. SNMP may be used to query the status of or set the values of network components. in a MIB is referenced by an object identifier (OID).9. the information on each device is organized in a virtual store called a Management Information Base (MIB). In the case of Cisco routers.3. A MIB is a hierarchical. the agent.2.1c 151 . Each piece of data. When requesting the value of an object. 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.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. numerical name. the router is always the agent.2. where the dots separate branches in a MIB . for more information consult [7]. values for the entire branch are returned.2. tree-like structure used to store a virtual database of network management information. An SNMP agent device maintains information and makes it accessible to managers.9.NET-SNMP is the successor to ucd-snmp.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. A free and usable implementation of an SNMP agent and manager may be obtained from the NETSNMP home page (http://net-snmp. SNMP is the transport protocol used to share and change information between MIBs.sourceforge. An in depth discussion of SNMP data organization is outside the scope of this guide.

DES 56-bit encryption was not supported in the versions of IOS used for preparation of this section (12. 152 Version 1. However. Consequently strong security was left out of SNMPv2c. SNMPv2 relies on the same mechanism for access control to the MIB.0 and later support all three versions of SNMP. SNMPv3 defines three levels of security. conflict arose between competing proposal advocates and no security standard was agreed upon. In the late 1990s. However. SNMP Vulnerability In early 2002. SNMPv1 uses a community string to limit access to the MIB. 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.3 supports SNMPv1 and SNMPv2c. Proposals were put forth to integrate security (as well as more functionality) into later versions of the protocol. Cisco IOS supports a large number of SNMP-related commands. 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. SNMPv2c and SNMPv3.0(7) and 12. it was originally intended to be a short-term solution for (remotely) managing networks. Unfortunately.0(5)). SNMPv1 and SNMPv2c have weak security. consult the Cisco security advisory “Malformed SNMP Message-Handling Vulnerabilities” [9]. IOS versions 12. disable SNMP altogether.0 supports all three security levels. since it was the only network management protocol available at the time. it became widely used. For more information. SNMP Security When SNMPv1 was developed. They are described in the table below. As such. SNMPv3 was developed specifically with strong security in mind.Router Security Configuration Guide problems. This string is sent across the network in clear text. This section will give a brief overview of SNMP security and will detail how to enable SNMP more securely. or take other protective measures. those that do not have a direct impact on security are not covered. There are currently three versions of SNMP: SNMPv1. it was developed quickly and strong security was not a requirement. IOS version 11.1c .

While there is some mention of enhanced security options (for SNMPv2c) in the Cisco documentation. 0 sent. End with CNTL/Z East(config)# snmp-server community publicstring East(config)# snmp-server host 14. It is a good idea to run the show snmp command to display the SNMP status and statistics. there are some basic commands you must run to enable SNMP.2.6 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 14.2.0. on Cisco IOS version 11. Unfortunately. Then an access control list (see Section 4.5. these commands have been disabled. It is a good idea to disable the public community string. one per line. anybody who knows this community string has access to essentially the entire MIB.1).6. So.1c 153 .6. 0 dropped.3 (which implements SNMPv1 and SNMPv2c). SNMPv3 A Cisco router capable of running SNMPv3 allows for more security measures to be applied. East# config t Enter configuration commands. In order to enable SNMP a default community string must be set.6.162. East# Running these basic commands by themselves is not very secure. 0/10.Advanced Security Services Configuring SNMP . SNMPv3 has been implemented and provides more security features.3) needs to be created to limit machine access to the router Version 1. SNMP logging must also be enabled (see section 4. as shown below.Getting Started In both IOS versions 11 and 12. there is no other alternative when enabling SNMP. in version 12. This string is stored on the router in clear text and will be sent across the network in the clear. The rest of this section focuses on SNMPv3. However.

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

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

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

ifTable: Central# config t Enter configuration commands. . etc. gauge.1 every 30 second(s) Taking delta samples. However. Central# Version 1. 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. All monitored objects must include an instance number of the monitored variable.13. last value was 3 Rising threshold is 40.1 30 delta rising-threshold 40 1 falling-threshold 0 owner rscg Central(config)# exit Central# show rmon alarms Alarm 1 is active. For example. there are several limitations on the type of SNMP variables RMON is capable of monitoring.1c 157 . owned by config .1 notation. Alarms may define any SNMP MIB variable that has an elementary data type such as integer. the following command defines an alarm configured on a member of the MIB II interfaces table. It is acceptable to use the Object Identifier (OID) or the qualified variable name that resolves to its OID. Variables included in the SNMP table format will have an instance number equivalent to the entry number of the table. one per line. End with CNTL/Z. All other elementary data variables should have an instance number of ‘0’. timeticks. counter.Advanced Security Services and the falling thresholds of an alarm are dependent on the previous configuration of an associated event. The MIB object monitored must also resolve to an ASN. assigned to event 0 On startup enable rising or falling alarm Alarm 2 is active. owned by rscg Monitors ifEntry.13. RMON alarms provide an excellent tool for monitoring the network interfaces supported by the router. Use the rmon event command to provide a description of an event and specifies whether a message is logged or a trap is generated. Central(config)# rmon alarm 1 ifEntry. Use the rmon alarm command to designate the actual MIB variable monitored on the Cisco router. An important requirement that is easily overlooked is the instance number of the monitored variable. assigned to event 1 Falling threshold is 0.

This section also presents information about backing out of an upgrade. RELEASE (fc1) Copyright (c) 1986-1998 by cisco Systems. booted via flash cisco 3640 (R4700) processor with 28672K/4096K bytes of memory.T1". If the amount of traffic generated by RMON seems to be too high. It is important to note that most Cisco updates can only be accomplished by replacing the IOS software running on the router.5.Router Security Configuration Guide The interface entry. input and output discarded packets. System image file is "flash:c3640-i-mz. An upgrade can be beneficial for security.113-4.g. To determine the current software release running on a router. Inc. . It is important to know the router model and memory sizes before attempting to obtain a software upgrade. . 30 seconds to 60 seconds). the number of inbound packets discarded. To compute the total RAM on the router. The agent monitors increases of forty discarded packets or more starting from the last value sampled. there is no facility for amending or patching installed IOS software. A router’s RMON agent can be very useful for monitoring the number of checksum. and flash memory size. Central> The underlined portions of the transcript are the software version. Version 11. but if done improperly it can leave a router vulnerable. and identify the version and memory size as shown below. etc. RAM size. use the command show version. simply add the two parts of the RAM size rating: this router has 32MB of RAM. . . then change the sampling period to a longer time (e. unknown or unsupported protocols.3(4)T1.13.1c . RMON may be very data intensive depending on the number of monitored variables and the length of the sampling period. respectively. ifEntry. Central> show version IOS(tm) 3600 Software (C3640-I-M). 4. 158 Version 1. .5. 8192K bytes of processor board System flash (Read/Write) . input and output errors. 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. router model. Performing Cisco IOS Software Updates This sub-section outlines the motivations and procedures for upgrading the system software on a Cisco router. identifies variable ifInDiscards.1.

Third. and bug history. Weigh the benefits of upgrading against the risks before you start. you can download updates from the Software Center on the Cisco web site. It is very important to read the release notes for a new release and test it carefully before installing it for operational use. and again afterwards. To fix known vulnerabilities – when security vulnerabilities are found in Cisco IOS products. either by implementing new features or by reducing available free memory. The logistics of purchasing updates is beyond the scope of this document. If you have a maintenance agreement with Cisco. If the performance benefit is greater than the cost of upgrading. Deciding which update to pick is a complex topic. 2. you will be given the length of the release in bytes. one solution may be to upgrade to a later edition of the IOS software. release status. To support new features – Cisco has added new operational and security features to each new IOS release. Software updates entail substantial costs. then it makes sense to upgrade. Whenever you download Cisco IOS software (often called an IOS “image”). some features may not work in a newer release. cost. then do the upgrade. you must take many factors into account: feature availability. To improve performance – you might need an upgrade to support new hardware or hardware features. for the IOS image you’ve Version 1. For more information about Cisco IOS release types. 1. If the performance of your router is critical. During the software selection and download sequence at Cisco’s web site. Obtaining Updates Cisco makes software updates available through a variety of purchase and maintenance mechanisms. the router must be out of service for at least a short time during the installation process. Second. depending on router model and other factors. to ensure that the new software can fully support the router functions your network needs. router memory size. see Section 8. First. 3. which will include the length. it is best to check the length after downloading. Print the summary web page. then measure the performance before upgrading. they might be broken or simply unsupported.3.1c 159 . the minimum downtime will range from at least a minute to several minutes.Advanced Security Services Motivations for Updating Router Software Installing an IOS update entails inconvenience and the risk of disruption of service. or to enforce your local security policy. If you need one or more of these features to support your network. be prepared to back out if the performance has suffered. a new release may degrade performance. The list below describes some good reasons for installing an update.

After downloading the IOS binary file. 3. you will need to have an SSH server running on a host in your network.3 or earlier. and may impose much longer downtime (up to half an hour if things go wrong and you have to back out). Schedule your downtime. it is not a secure service. Read the entire upgrade procedure. If possible. and inform the user community as needed. Most newer IOS releases support SCP. Every Cisco IOS release has minimum memory requirements. use SCP or FTP for performing Cisco upgrades. because many Cisco routers are configured with just enough memory to run the IOS version pre-installed at the time of purchase. and should not be running in a secure network. most earlier releases do not.Router Security Configuration Guide selected. use the md5sum command. 4. Check your TFTP. RCP. or FTP configuration. or a similar checksum utility.1c .) 2.) While TFTP is supported by all IOS versions. Do not install an update unless the router to be upgraded satisfies the memory requirements for both RAM and Flash. If the length or hash of your file differ from the summary page. discard the file and download it again. Cisco routers have two fundamental kinds of memory: RAM and Flash. then FTP will probably not be available. Be sure that you are familiar with all the IOS commands involved. Copy the new Cisco IOS software into the server’s download directory. it is prudent to configure operational routers with as much memory as they can hold. 1.0 supports FTP. check the length against the printed page. Use the commands show version and show flash to check the amount of memory your router has. consult [12]. Cisco IOS 12. For more information. When possible. a secure file copy protocol based on SSH. Installing an update imposes a minimum downtime. Before You Perform the Update Check all the items below before installing a new IOS image on your router. 160 Version 1. (If the router to be upgraded is running IOS 11. Schedule your upgrade ahead of time. (Often. Review the entire procedure before you start. to check the hash value. Make sure that the TFTP or FTP server is correctly set up for both upload and download. below. then disable it again. Ensure that you have enough memory. a major upgrade will require more memory. Enable TFTP only for the update sequence. To use SCP. The summary page will also include the MD5 hash value for the IOS image. Router software updates are usually performed using TFTP or FTP.

Update Procedure This section presents a suggested sequence of steps for installing Cisco IOS software. the router will attempt to log in as user “anonymous”). then disable the outside network interfaces using the shutdown command. 2. If this step fails. set an FTP username and password that will work on your server (otherwise. Central# config t Central(config)# interface eth 0/0 Central(config-if)# shutdown Central(config-if)# end Version 1. confirm the current IOS and boot version. The backup phase. The console will show important status messages in the later steps of the installation that would not be visible otherwise.Advanced Security Services If possible. If you are planning to use FTP. make a record of them. If a redundant router or a hot spare is available. For example: Central(config)# ip ftp username uploader Central(config)# ip ftp password Y3ti4ttack! Central(config)# end 1.1c 161 . If the router to be upgraded is a border router. and back up the current IOS software. do not proceed. Enable privileges. steps 56. it is safest to replace a router and take it offline for update. The sequence has three phases: backup. take advantage of that to perform the update without disrupting service. by following it you can avoid mishaps. Log in on the router console. The test phase. Check the current IOS version number and flash contents with the commands show version and show flash. The sequence is very conservative. and ensure that you can restore your previous IOS version if necessary. steps 1-3. It is best to perform router updates from the system console rather than from a network login. 0. involves copying the running IOS software and configuration onto the TFTP server host for safekeeping. 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. followed by a console transcript of a successful update. The steps are described below. Shut down external interfaces. and test. abandon the update and check the server configuration before trying again. The install phase. involves loading the new software. involves checking that the new software is running the old configuration successfully. Copy the router’s current IOS image to the server using the copy command as shown below. step 4. install.

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

If you shut down your router’s external interfaces in step 2.T1 Destination file name [c3640-i-mz.bak Verifying checksum for 'c3640-i-mz.0(5).T1 [3208612 bytes used. South>show flash System flash directory: File Length Name/status 1 3208548 c3640-i-mz.T1' from Flash to server as 'c3640-i-mz-113-4.9.2. then bring them back up now using the command no shutdown. South> show version Cisco Internetwork Operating System Software IOS(tm) 3600 Software (C3640-I-M).T1 [3208612 bytes used.3(4)T1. RELEASE SOFTWARE (fc1) .113-4.T1. Version 11.Advanced Security Services 6. 8388608 total] Address or name of remote host [14. If the second command in step 5 showed that they did not come back up. 5179996 available. Note that. South(config)# interface ethernet0/1 South(config-if)# shutdown South(config-if)# exit South(config)# exit South# Version 1.113-4. . OK Copy 'c3640-i-mz. Bring up external interfaces. if necessary. Transcript of a Successful Update Procedure The recorded transcript below shows an upgrade of a Cisco 3640 router from IOS 11.113-4. this procedure may take about 5-20 minutes.3(4) to 12. End with CNTL/Z..2. they should have come back up as part of the reload in step 4. Consult the release notes for the particular router for details.. Central# config t Central(config)# interface eth 0/0 Central(config)# no shutdown Central(config)# end Depending on network speed and router model.6]? 14.bak'? [yes/no]yes !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Upload to server done Flash device copy took 00:00:19 [hh:mm:ss] South# config t Enter configuration commands. 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.113-4. 5179996 available.1c 163 . one per line.T1' (file # 1).6 Source file name? c3640-i-mz.T1.T1]? c3640-i-mz-113-4. for some older Cisco router models.113-4. additional hardware-specific steps may be needed.9.113-4.

bin from 14..Router Security Configuration Guide South# copy running-config tftp Remote host []? 14.255]? 14. 164 Version 1. entry point: 0x80008000..9.bin]? c3640-ik2o3s-mz_1205_T1.erased Loading c3640-ik2o3s-mz_120-5_T1.. Inc.bin from Save? [yes/no]: no Proceed with reload? [confirm] y %SYS-5-RELOAD: Reload requested System Bootstrap.. eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee ..bin' on 14.9...6 (via Ethernet0/0): !!!!!!!!!!! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! [OK .9.2.2. C3600 processor with 32768 Kbytes of main memory Main memory is configured to 64 bit mode with parity disabled program load complete.6? [confirm] Building configuration. 8388608 total] Address or name of remote host [255..9.T1 [3208612 bytes used.9.bak on host 14.255.2. Version 11.113-4.2.bak !! [OK] South# copy tftp flash System flash directory: File Length Name/status 1 3208548 c3640-i-mz..9.2.7656076/8388608 bytes] Verifying checksum.6 Source file name? c3640-ik2o3s-mz_120-5_T1. size: 0x74d170 Self decompressing the image : ############################################# [OK] .bin' into Flash WITH erase? [yes/no]yes Erasing device.255. OK (0xDC3B) Flash device copy took 00:00:50 [hh:mm:ss] South# reload System configuration has been modified.bak Write file south-config.2. Are you sure you want to erase? [confirm] Copy 'c3640-ik2o3s-mz_120-5_T1.. EARLY DEPLOYMENT SOFTWARE (fc1) Copyright (c) 1998 by cisco Systems..bin Destination file name [c3640-ik2o3s-mz_120-5_T1.1(19)AA. Writing south-config.bin Accessing file 'c3640-ik2o3s-mz_120-5_T1. South> South> show ip interface brief Interface IP-Address OK? Method Status Protocol Ethernet0/0 14.2.2. 5179996 available.6 (via Ethernet0/0): ! [OK] Erase flash device before writing? [confirm] Flash contains files. Loading c3640-ik2o3s-mz_120-5_T1. ..bin' from server as 'c3640-ik2o3s-mz_120-5_T1.1c .64 YES NVRAM up up Ethernet0/1 14.6 Name of configuration file to write [south-confg]? south-config.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.

many default settings differ between various IOS releases. Because TFTP provides no security.3). Also. First. some newer versions offer services not present in older versions (see Section 8.0 . you may be exposing your encrypted passwords to disclosure. use a different name than you used during the upgrade procedure. Also. Note that. always use it instead of TFTP. your stored configuration might not work correctly when you fall back to the older version. The simplest approach to mitigating this risk is to change the enable secret immediately after installation (see Section 4. load the backup copy of the old IOS software. but the simplest is to ensure that the TFTP traffic does not traverse hostile networks. Central# Central# . if you follow the installation procedure outlined above. . There are several approaches to doing this. Simply follow the procedure described above. . it is critical that you protect the TFTP transaction and server from potential attackers. copy tftp flash reload ! Optional. you transmit a copy of your router configuration to a TFTP server. from 11. restore the backup copy of the configuration that you saved during the upgrade procedure step 1. Version 1. Third. this section attempts to address them. do not leave TFTP enabled on your host.Advanced Security Services Current configuration: ! version 12. In that case. . Central# .0). In step 4. Second. Central# . whenever you make any kind of backup copy of a router configuration.1) or to use a centralized authentication server (see Section 4.g.2 to 12. South# exit Backing Out an Update If functional testing reveals a problem with your router after an upgrade.1c 165 . if you have upgraded from one IOS major version to another (e. If the router supports SCP. restore old configuration copy tftp running-config Additional Security Concerns There are several security issues surrounding upgrades. you may need to return to your old IOS version. always turn it off immediately after you finish the installation procedure. Some of these settings can affect your router’s security.6). . In step 3. starting with step 2.

and they offer a great deal of visibility into the operation of the router. The ones listed here are commonly used for simple troubleshooting. keep input/output statistics. or when a network is under attack. Diagnosing and Debugging Router Operation Effective logging and SNMP help an administrator to stay aware of their routers’ status and operational condition. even on the simplest Cisco routers. Testing and validation are covered in Section 6. Router Status and Configuration Commands Each of the items below describes a single status query. and the log messages.3 and later. see [2] and [7]. an unauthorized attempt to telnet to the router itself has been logged by access list 131. These commands typically produce a correspondingly great deal of output. for a discussion of some other useful ones. 1. identify sources and causes.5. • Debugging commands – Virtually every IOS facility and protocol has associated debugging commands.5. so the first few messages put into the log reflect startup activity. Cisco IOS diagnostic and debug facilities can be used to get vital information. When a problem occurs. so recent messages are at the end of the listing.2.6. so use them sparingly. The buffered log messages are cleared when the router reboots. There are IOS commands to display these statistics that can be used to detect problems. and some is particular to each interface. This short sub-section describes some of the most useful techniques for IOS 11.Router Security Configuration Guide 4. There are literally hundreds of such queries available. These commands can also be used to help verify that security measures are in force. some of this information is global to the whole router. Viewing the current log – To view the current buffered log messages. The output consists of two parts: a summary of the current logging configuration. consult Section 4. use the command show logging. and validate repairs. East# show logging 166 Version 1. and some other facilities. For more discussion of logging. The techniques fall into three groups: • Router status and configuration commands – These commands display information about the settings and tables held by the router. Techniques for troubleshooting and debugging routers could (and do) fill entire books. In the example below.1c . • Router throughput and traffic commands – Each interface. and are useful for understanding a Cisco router’s disposition in a typical TCP/IP network. The messages are shown in the order they occurred.

9.0 is directly connected.0.2.0/16 is directly connected. In a router using NTP.0. .0.1. 2d18h.250 to network 0. Figure 4-11 shows how to interpret the output of show ip route. 33 message lines logged Logging to 14.0.0/8 [110/12] via 14.Advanced Security Services Syslog logging: enabled (0 messages dropped.0. see Section 4.0. 1 packet East# Note: log messages should always include the time of the event.1.1.0. Ethernet 0/0 Route codes Destination gateways Figure 4-11: Interpreting a Route Table Listing Version 1. this list may be very large.2.17.0/0 [110/3] via 14. or were added by a configured routing protocol (codes R. use the command show ip route.0(23).250.1c 167 . Ethernet0/1 0.0 O IA O IA O C O E2 C R O*E2 7. Ethernet0/0 7.0 [120/1] via 14.0 [110/10] via 14.2. changed state to up . 56 message lines logged Trap logging: level informational. Ethernet0/1 14. 1d01h. or were added as static routes (code S). 32 message lines logged Buffer logging: level debugging. 1d13h.1. Mar 3 12:51:52 EST: %SEC-6-IPACCESSLOGP: list 131 denied tcp 0 flushes) Console logging: level debugging. Ethernet0/0 14.1.101. Viewing the current route table – To view the current route table. A very important part of reviewing the route table is checking the route codes and checking the destination gateway. Ethernet0/0 14.0. Gateway of last resort is 14. Each route code identifies how one route joined the table.4). 33 message lines logged Log Buffer (16000 bytes): 00:00:17: %LINK-3-UPDOWN: Interface Ethernet0. because the messages are generated before NTP has synchronized. 2. Ethernet0/0 14.250. Depending on the size of the network and the kinds of routing protocols used. Check the route codes to make sure that all the routes joined the table either directly (code C). and others. 00:01:05.250(47746) -> 0. Ethernet0/0 172.1. 56 message lines logged Monitor logging: level debugging. the destination gateway is simply the next hop on that route. O.0/8 [110/14] via 14. the first few log messages will include the time since boot instead of the correct time. 2d18h. Note that the route table listing on an operational router will often be much longer than this sample. [110/11] via 14. 2d19h.

2.20 10 Internet The output is different for each kind of protocol.2.1. To display the ARP table. 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 14.3680 0010.250 53 . and unauthorized devices on a network segment can often be detected by their presence in a router’s address resolution (ARP) table.127a 0050.20 Eth0/0 14. 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.2. Central# Hardware Addr 0004. The command show ip protocol summary gives a quick overview. Viewing the logged in users – The command show users displays a list of users that are currently logged in.9.250 1 FULL/DR 00:00:38 14. there is one user logged in at the console. see Section 4.2. Central# show arp Protocol Address Age(min) Internet 14.1. In the example output below. and two are logged in over the network.2.4 for some recommendations.20 1 FULL/DR 00:00:33 14.6 6.9. use the command show arp. All of the individual routing protocols also have extensive status commands.2.1c .1.Router Security Configuration Guide 3.64 43 Internet 14. mis-connected devices.2.1.0f03.f3f6 0010.1.20 idle 00:00:00 14. as in the example below.baa0 Type Interface ARPA Eth0/1 ARPA Eth0/0 ARPA Eth0/1 ARPA Eth0/0 5.2. Viewing the current ARP table – Extraneous devices. .6 57 Internet 14.7bb6.7bf9. Viewing host name and name lookup information – Cisco IOS uses two mechanisms for mapping between IP addresses and names: locally defined names.1. The example below shows the IP routing protocol summary and (abbreviated) output for a useful OSPF status command. and DNS. 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 14.250 Eth0/0 Central# 4. Locally defined names take 168 Version 1.acd5.

For a quick look.1.1. OK) south (perm. and typically five virtual line connections.1. In all cases. slightly abbreviated.Advanced Security Services precedence over DNS names.15.1. OK) Central# Age 4 ** 52 Type IP IP IP Address(es) 14.1 36269 14. OK) central (perm. use show line name number.15.6 514 14. the console. the telnet vty lines. but can provide valuable clues to the services that the router is actually providing.250 161 17 0. the listing will include both active and inactive interfaces.250 123 17 14.2. 9.1.250 6082 Central# In Out Stat TTY 0 0 1 0 0 0 1 0 0 0 1 0 0 0 10 132 Version show line aux 0. Viewing currently open UDP sockets – Use the command show ip socket to list the currently open UDP network service sockets on the router.0. Central# show ip interf brief Interface IP-Address OK? Ethernet0/0 14.0 520 14.250 520 17 14. Use the command show line to display a summary of lines available on a router (see Section YES Ethernet0/1 14.1 Host Flags east (perm.1.0 123 14.2.1. Central# show host Default domain is not set Name/address lookup uses domain service Name servers are 14.2.2. The output is a little cryptic.1. The example below shows the output for a router running fairly few services.0.9. for instance. use the command show ip interface brief.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 8.1c 169 .0.1.64 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. Use the command show host to display the DNS configuration and the list of locally defined names.4). To display the full status of a line.250 14.2. Viewing line status – Every Cisco router has at least one physical line connection.1. The example below shows the brief output format. 14. Central# show ip sockets Proto Remote Port Local Port 17 0.

1c . The resulting output will typically be fairly long. 170 Version 1. and also shorten downtime if the router loses its stored configuration. use the command show running. and review the hardcopy.5. it should be used with care and disabled when not in use. FTP is not supported. sending syslog messages to a Unix host (remote port 514). Viewing the current configuration – To view the current running IOS configuration. print it. Viewing currently running processes – Many IOS services and facilities run as separate IOS processes. these two configurations should be very similar. If the configurations are very large and complex. Because TFTP is insecure. 10. or on a monthly basis. using IOS 12. Normally.Router Security Configuration Guide The first line is the RIP route protocol service (local port 520). Use the command show process to list the running processes. The second line is the SNMP service to a host running an SNMP/RMON management tool (local port 161). Archive a copy of the configuration after any major change. The example below shows how to save an archive copy of a configuration to an FTP server.2. The third line is the network time service (NTP. Another way to get an archive copy of the running configuration is to use text logging features of Telnet and terminal emulation applications. one per line. Check for unwanted processes and services.0. This can help with problems.1 Destination filename [central-confg]? central-config. Central# config t Enter configuration commands. such as Unix diff or Windows fc. The output is usually quite long. To review a configuration in depth. use a file comparison tool.txt !! 5699 bytes copied in 12. 11. port 123). to highlight the differences. The fourth line is the logging client. but TFTP can be used for making archive copies in a very similar manner (see Section 4. 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 []? 14.3 and earlier.txt Writing central-config.9. save the command results to a file.716 secs (474 bytes/sec) Central# In IOS 11. To view the saved startup configuration (in NVRAM) use show startup.5). End with CNTL/Z.

991606 packets input.250/24 . 78 collisions. Central# clear counter Eth 0/0 Clear "show interface" counters on this interface [confirm]y Central# 2. This will display a comprehensive report on the traffic through all the interfaces. use the command show ip traffic. . The output is quite long.1. Viewing IP Protocol Statistics – To display a long listing of IP and related protocol traffic statistics.0 giants. line protocol is up Hardware is AmdP2. 0 no carrier 0 output buffer failures. 0 no buffer Received 800624 broadcasts.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. Last clearing of "show interface" counters 23:20:53 .0 runts. To view the traffic for a single interface. The example below shows the output format for a single Ethernet interface. Central# show interface eth 0/0 Ethernet0/0 is up. misconfiguration. 0 overrun. There are hundreds of traffic and processing status commands in Cisco IOS. 0 late collision. The example below shows how to clear the counters for a single interface. use the very Version 1. . Clearing the counters sets the traffic volume record back to zero for both input and output. 3. address is 0050. 0 ignored 0 input packets with dribble condition detected 480919 packets output. simply supply that interface name to the command. or component failure. 38371898 bytes. see [7] for more information about some of them. use the command show interface.0 throttles 0 input errors. 0 underruns 0 output errors.1c 171 . 1 interface resets 0 babbles. Viewing the network traffic on a per-interface basis – To view the total traffic for each interface. . 215 deferred 0 lost carrier. but can reveal certain classes of attacks. Understanding normal network and link traffic loads can be critical for identifying anomalous conditions that are indications of attacks. 0 frame.15. 0 output buffers swapped out Central# If traffic volume monitoring is important for a particular interface. Viewing SNMP Protocol Statistics – To display the SNMP messages statistics and configuration. . Most of these commands produce voluminous but clearly formatted output. 1. 103806395 bytes.7357.cbe0 Internet address is 14. 0 CRC. then clear the counters on a periodic basis.

2. 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. debugging can impose a substantial computational burden. or feature of the router. all activities of that protocol or feature will generate log messages at level 7. and should be used sparingly on operational routers.Router Security Configuration Guide simple command show snmp. are sent to all log sources configured to receive them. when generated. When debugging is enabled for a particular protocol or feature. Central# show users Line User *130 vty 0 rscg Central# config t Host(s) idle Idle Location 00:00:00 14. be sure to configure the buffered log and syslog for level 6 (informational).1c . Each debugging command is associated with a particular service. The example below shows the output for a router with a very modest amount of SNMP traffic. Router Debug Commands Cisco IOS offers a very extensive suite of debugging commands. facility.9. Also. Debug messages. and the network does not have an SNMP infrastructure deployed. If the output shows any SNMP traffic. Therefore.6 172 Version 1. then the router may have been subjected to an SNMP sweep by an attacker. The example below shows how to configure debugging and turn on debugging messages for ICMP. when using the debug messages for interactive troubleshooting. The number of messages generated by debugging can often be quite large.

References [1] Albritton. The section on “File Management” provides extensive information on downloading updates.1c 173 . “ping central” was performed on 14. McGraw-Hill. RFC 1305. [3] Ballew. including examining the configuration and operation of a Cisco router. An amazingly detailed book about IOS versions and the IOS release process.9. The sections on “Performing Basic System Management” and “Monitoring the Router and Network” include valuable advice on how to configure basic features and services. O’Reilly Associates. Managing IP Networks with Cisco Routers.0 Configuration Fundamentals. J. M. A practical introduction to the concepts and practices for using Cisco routers. the version supported by IOS 11 and 12.9. Cisco IOS Essentials.5. [4] Mills. Cisco IOS Releases: The Complete Reference. the Cisco IOS Debug Command Reference. [2] Cisco IOS 12. The specification for NTP version 3.6 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. [5] Coulibaly.2.. D.Advanced Security Services Enter configuration commands. An excellent introduction to basic IOS operations. 1995.2.7.M. 1997.9. Consult this book for information on upgrade paths and compatibility.2. End with CNTL/Z. 1999.6 Mar 3 17:02:13 EST: ICMP: echo reply sent.2. 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. 4. 2000. Cisco Press.250. one per line.6) Central# terminal monitor Central# debug ip icmp ICMP packet debugging is on Central# ! At this point. Version 1.159 EST: %SYS-5-CONFIG_I: Configured from console by rscg on vty0 (14. “Network Time Protocol (version 3)”. 1999. Cisco Press.M. dst 14.9. S. src 14.

and Wright. and Smith. Cisco Press. Feb 2002. Prentice-Hall. available at: http://www. E.Best Practices White Paper”. 1999. with some configuration suggestions.. products_security_advisory09186a00800945b4. Cisco Technical White Papers. available at: http://www. 174 Version 1. This Cisco guide for Internet Service Providers includes a good discussion of IOS upgrades. including good coverage of the SNMP basics and SNMPv3 security features. 1996.html This magazine article from 1999 gives a clear and readable overview of NTP. P.. [11] “Cisco Network Time Protocol . Understanding SNMP MIBs. A detailed exploration of the SNMP management information base. It also shows how to set up an IOS router as an NTP master. [12] Stewart.. SANS Institute.1c . and Perkins.html This note offers guidance on NTP hierarchy design and tracking performance especially for initial set-up of a new router. Network Computing. January 1999. An in-depth study of SNMPv3 and its use..shtml This Cisco security advisory provides detailed information for dealing with the 2002 SNMP vulnerability on Cisco IOS devices. available at: http://www. Securing Cisco Routers: Step-by-Step. Cisco ISP J. as well as examples of router status and diagnostic commands.. G. 2002. Prentice Hall. B. D. April 2002. Cisco Security Advisory. Cisco Systems. including both standard and vendor-specific structures. A Practical Guide to SNMPv3 and Network Management. [9] “Malformed SNMP Message-Handling Vulnerabilities”.cisco. Cisco Systems. 1st edition. [8] Greene. A very specific guide to configuring many IOS features [10] Shipley.networkcomputing. Includes guidance on using SCP for updates. “Getting in Sync: A Look at NTP”. [7] McGinnis. J. May 2002.Router Security Configuration Guide [6] Zeltserman.

and Kerberos security services provide the facilities required for AAA.Advanced Security Services 4. Security for Router Network Access Services Security for Network Access Services deals primarily with controlling remote users who are accessing local resources. and line passwords. authorization. AAA allows you to employ.6. authorization. When AAA services are enabled on a Cisco router. enable. Overview. 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. when there a large number of routers so that centralized administration becomes advantageous. This means that you can no longer access the commands to configure the older protocols (including login local and login commands).1. and many releases also support Kerberos. Where the older access control mechanisms dealt almost solely with user authentication. TACACS+. All Cisco IOS releases support the RADIUS and TACACS+ network security services. Cisco provides this security with their authentication. 4. except Kerberos does not accept accounting records. The sub-section below dealing with dial-in users will give an introduction to controlling remote users accessing network resources. selectively. Configuration is performed by using method lists as described further below. when flexible authorization capabilities are required.6. Basic Concepts. 2. AAA is the mechanism Cisco recommends for access control. or 3. Centralized authentication. and Support Mechanisms Cisco’s authentication. An Internet Service Provider would be a good example of this. and provides additional capabilities for auditing of network service usage. AAA allows you to base authentication decisions on the router’s local user database. when accounting is required. This allows for easier management of user accounts and privileges. RADIUS. By using AAA along with a security server you can control access to routers and other network services from a centralized location. Version 1.1c 175 . In addition to network security services. AAA is designed to allow the administrator to configure its services globally or by line and interface. authorization. Three conditions make using network security servers a good choice: 1. the older forms of access control are disabled. and accounting (AAA) services. AAA also has the ability to control each user’s access to resources and provides additional accounting capabilities beyond the router’s logging facilities. both network security services and security information local to the router. and accounting are particularly important when your organization has many routers and other network devices to manage. and accounting services provide critical security functions necessary for providing remote access to routers and network resources.

2.1c . Enable AAA (new-model). 176 Version 1. 4. For more information.2 demonstrates how to set up AAA authentication. it is best to use capable and well-managed network security services as your primary AAA mechanisms. The authentication method list defines the types of authentication to be performed and the sequence in which to apply them. Define one or more method lists for AAA authentication. and configure local user or line password support only as fallback mechanisms for when the network security services are unavailable. and supported token technologies.2. authentication controls the ability of a user or another network component to access a network device or service. 3. The default method list will be automatically applied to all the lines and interfaces for which a named list was not applied. if no local usernames are defined then remote administration (via Telnet.1). TACACS+. Authentication Authentication is the mechanism for identifying users before allowing access to network components or services.6. 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). to use security server protocols or backup authentication methods you must use AAA authentication. Configuring AAA authentication entails four basic steps: 1. SSH. see the sub-section on “Method Lists” below. in this case. line passwords. local username database. Although authentication can be configured without using AAA (see Section 4. enable passwords and none. The following sections will discuss the three main faculties provided by AAA and their supporting concepts.6. In other words. AAA authentication provides the means for identifying users through login/password dialogs. When AAA authentication has not been set up the default will use the local username information. AAA authentication is set up using method lists. etc) will not be possible. Apply the method lists to a particular interface or line (optional). For AAA authentication the available methods are RADIUS. Configure security server network parameters. Examples in this section will use a subset of the main network diagram as shown in the "Putting It Together" sub-section in 4. challenge/response mechanisms. Section 4. Therefore. Named lists must be applied to the appropriate lines and interfaces. but console access will be allowed.Router Security Configuration Guide Note: When using the local user database instead of a network security server. Kerberos. AAA is very limited in its authorization capabilities and provides no mechanism for accounting.

if-authenticated and local. must be applied to the interface or line to be invoked. authorization for each service. as discussed in Section 5. If no authorization list is defined for the authorization type then no authorization will occur for that type. Authorization works by creating a list of attributes which describe what the user is allowed to do. define security servers. • reverse-access – controls access to all reverse access connections like reverse Telnet. The RADIUS and TACACS+ security servers. and ARA protocols. authorization can only be configured using AAA. To prevent passwords from being exposed in the clear you must use a secure remote administration with a protocol like SSH or IPSec. Default method lists are applied to all lines and interfaces for that particular authorization type. Authorization includes one-time authorization. Authorization method lists can include RADIUS and TACACS+ security protocols along with Kerberos Instance Maps. Authorization lists are specific to the authorization type which is being defined. PPP NCPs. as described in Section 4.4.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. Authorization is the method used to describe what a user has the right to do once they are authenticated to the router.6. • network – enables authorization for all network related services like: PPP. But named method lists. use attribute-value pairs to define a user's rights.1c 177 . if a user has been authenticated when logging into the console line then authorization will not be used (even if configured). Authorization Authorization controls access to system resources. then the security server database will be used to control access to various network components and services as defined by the stored attributes. if-authenticated. Version 1. and authorization for each user. After a user logs in and has been identified by authentication. and define the rights for each user. AAA authorization types are: • exec – which controls the users ability to run an EXEC shell. AAA does not protect the confidentiality of the password during the trip from the remote administration host (e. and local methods. PC on the administrator’s desk) to the router. • commands <level> – which controls access to all the commands at the specified privilege level. Prerequisites for AAA authorization are: enable AAA services. SLIP. are very limited. other than “default”. Additionally.) As with authentication. configure AAA authentication (since authorization relies on authentication's output). method lists define what authorization protocols will be used and in what order.g. There is a special case for the console line. (The last two methods.

but RADIUS does not support command or system. number of packets and bytes. etc. Accounting AAA accounting is used for logging and tracking the activities of users (people or other network components) using a network resource. Cisco IOS does not support command acccounting with RADIUS. the usage start and stop times. If the list was a named list then it must be applied to the appropriate lines and interfaces. • system – Provides information about system-level events. The list will define the list of accounting methods for the indicated accounting type. Routers send their accounting records to the security server for storage. 178 Version 1. start and stop times. Account records are generated based upon the level of the command not the level of the user.6. and the user. and ARAP protocols. For an accounting type. These logs can be used for network management. As with authentication and authorization. date. • connection accounting – Provides information about all outbound connections made from the network access server.Router Security Configuration Guide Section 4. There are several types of accounting which can be enabled and configured separately: exec. Information in an accounting record includes the user’s identity. command. AAA accounting can only use the TACACS+ or RADIUS security servers for record logging. network. Accounting records will include the command. • network accounting – Provides information for PPP. The information includes the username. 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.2 shows an example of configuring AAA authorization. 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.1c . system. This would include information like system reboots. The information includes the number of packets and bytes. • EXEC accounting – Provides information about user EXEC sessions on the router. SLIP. time. refer to the “Configuring Authorization” chapter in the IOS Security Configuration Guide [1]. you configure AAA accounting by defining a list of accounting methods. date. For more detailed information about configuring authorization using AAA. This option will apply accounting to all commands issued at the specified privilege level. resource usage tracking. IP address of access server. • command accounting – This applies to commands which are entered in an EXEC shell. rlogin. accounting being turned on or off. security analysis. This includes telnet. and reporting. and the command that was executed. All types are supported by TACACS+. and telephone number the call originated from for dial-in users. connection.

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

this section focuses on using AAA for the security of remote administration. For detailed information about server groups. there is directly controlling authentication to the router and then there are commands for providing information about the authentication process. ! AAA method list syntax for IOS 12.3 and 12.6.1 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. First.0. and the FreeRadius server was used for additional RADIUS testing.2. The rest of this section uses the IOS 12.0 syntax.6. The name of the first list is “remoteauth”. Use of RADIUS or TACACS+ requires the keyword group.3 briefly covers a dial-in user example. AAA configuration is a broad subject. the syntax for method lists changed slightly with the introduction of security server groups. Authentication The AAA authentication commands can be grouped into two areas which correspond to how they are applied. 4. Cisco's ACS Version 2. for named and default method lists.1 or later. simply insert the keyword “group” in front of the words “radius” or “tacacs+” when attempting to apply any of the examples. In order to use Cisco's AAA mechanisms you must first enable AAA services. Section 4.3 was used for testing RADIUS and TACACS+ security servers. The four authentication commands used for controlling access to a router are: 180 Version 1. Section 4. and accounting mechanisms and how method lists are used to define the security protocol to use for a service.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.Router Security Configuration Guide The following example shows the syntax.4 describes security server protocols in more detail.1c . 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. in IOS 11. consult the IOS 12.6. ! AAA method list syntax for IOS 12. if you are using IOS 12. there is a small example in the next sub-section. Router Access Control The previous section introduced authentication.1 and later.1 Security Configuration Guide. authorization. as shown below. This section will cover details of configuring AAA for controlling access to the router.

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

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. 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.1c . Use the enable secret password to protect all higher privilege levels. and should use the local user database. then any additional method which requires a username will automatically fail. authentication will use the local database on all lines. decide whether to use usernames and passwords (preferred) or to use just a password (highly discouraged). authorization should be used to control the larger set of administrators. 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.Router Security Configuration Guide Here is an example of setting up local username and password and AAA default login authentication parameters. To avoid being locked out of your router. make sure you add an administrator account to the local username name database before enabling AAA. In a more complex scenario where a more limited set of administrators have access to the console line. For accounting purposes you should use the methods which allow for usernames and assign each administrator a distinct username. including VTY lines. For the named method list which will primarily use the security server. The following is a recommended configuration for using a RADIUS security server and the local user database as described above. When designing your method lists. first create the default list. 182 Version 1. The default list should be for the limited set of administrators. If the first method in the list is line or enable. should apply to the console line only. Important: when AAA is turned on. 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. The second list should be a named method list and should be applied to the appropriate lines. 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. to allow additional administrators remote access to the router. then by default.

Of the four authorization types. This section will concentrate on exec and command authorization. AAA authorization is currently of limited use for controlling access to routers beyond the standard authentication mechanisms. First. Second.6. The no form of this command will turn off authorization on configuration commands in the EXEC mode. 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.1c 183 . if the router is used for dial in access. For scenario 1 there would be additional considerations as described in the Dial-In Users section. and Section 4. There are two primary scenarios where authorization is useful. arap. the other two (network and reverse-access) primarily deal with dial-in and dial-out access control and apply to interfaces. etc. • 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). • (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). Another network type. In scenario 2. The router Version 1. 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. exec and command deal with router access control and apply to lines.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. Scenario 1 – Router with dial-in users. and who can access and configure the router. and will not be covered. is also applied to lines.3 on Dial-In Users provides an overview of network and reverse-access authorization. authorization is useful for controlling who can access network services. authorization can control different administrators who have access to different privilege levels on the router.

refer to the "TACACS+ Attribute-Value Pairs" appendix of [1]. associated with the user. refer to the "RADIUS Attributes" appendix of [1]. • show accounting {system | network | exec | commands level} {start-stop | wait-start | stop-only} tacacs+ command can 184 Version 1. 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. Periodic generates more accounting records than newinfo since it will also include interim reports on actions in progress. RADIUS and TACACS+. AAA authorization does not apply to the console line. 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. For a list of supported RADIUS attributes. Also. But there are no useful parameters to set for access to the router from lines (an exception would be for dial-in access).1). There is one very important exception to this. Even if a named method list is created and applied to the console line authorization will be ignored.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. • (line): accounting {arap | commands level | connection | exec} [default | list-name] can be used to apply different accounting services and levels to different lines. attributes are defined on the security server. which are stored in a database on the security server. The local database is populated using the username command. Important: do not use the username name privilege level command since the password will be weakly protected. RADIUS and TACACS+ authorization both define specific rights for users by processing attributes. and sent to the network access server where they are applied to the user's connection. • aaa accounting suppress null-username command prevents accounting records from being generated for those users who do not have usernames associated with them. For a list of supported TACACS+ A-V pairs. For both.1c . Protect higher levels on the router using the enable secret command (see Section 4. (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.

For command accounting stop-only will provide the necessary coverage and will greatly reduce the number of accounting records. If your security policy calls for keeping a record of every router command. and will actually be terminated. This is not a configuration command but is worth mention. There are two basic scenarios for accounting depending upon which security server is in use. • stop-only sends an accounting record at the end user process which is of an accountable type. In this case the user process can not continue. then you must use TACACS+ accounting. As mentioned earlier Cisco's RADIUS implementation does not support system and command accounting. This is all done in the background and the user process will continue regardless of the outcome of the accounting attempt. • none specifies that no accounting records will be generated for a particular accounting type. 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.1c 185 .Advanced Security Services be used to show active connection information. • wait-start accounting sends an accounting record at the start and stop of each specified type. if the start accounting record can not be recorded. 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. 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. 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. 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.

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


6. Configuring dial-in services is far too complex a subject to be dealt with in depth in this guide. RADIUS was designed with Dial In access control in mind and the accounting features are very flexible along these lines. However Cisco's RADIUS 192 Version 1. The following descriptions will only discuss items which are different from the Putting It Together examples in the previous section. RADIUS can be used to provide authentication. particularly the “Dial Solutions Configuration Guide”. Security Server Protocols In Cisco routers and network access servers. and Kerberos. TACACS+. If a TACACS+ server was used in this example instead of the RADIUS server then system accounting would have also been specified. for more details.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. Command level accounting could have been applied as well but would probably not be needed here.1c . accounting was added for recording network events. Also the AAA authorization and accounting default commands for network will also apply to the ppp traffic as it traverses the line. AAA authorization for exec and network was added to separate the privileges for network users and router administrators. Cisco supported security servers are RADIUS. 4.4). RADIUS protects the communications using a shared secret. In addition. 3) the asynchronous interface configuration needs completed (if the aux port is not used as an asynchronous interface disable it see Section 4. RADIUS is a distributed client/server based architecture used to pass security information between access points and a centralized server. This section only provides one example for a possible network access server configuration. and accounting services. 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). The asynchronous interface contains the commands necessary for configuring AAA authentication for the ppp protocol. Security servers are important to Cisco network gear when centralized administration is required or when authorization and accounting services are needed. RADIUS Remote Authentication Dial In User Service (RADIUS) is an IETF proposed standard (RFC 2865) for securing network components.1. 2) the terminal server lines have not been configured (and will need to have the remoteacc accounting list applied) and.4. authorization. Consult the Cisco IOS documentation.

2. • radius-server key string sets the RADIUS server shared encryption key. These attributes can be configured to allow/deny access to various router and network services. use the command aaa server group. if you have defined it as described in Section 4. The secrecy and quality of this key is critical to the security of your RADIUS installation.2.6. so make it longer than a typical password. users never have to type this string. Use the auth-port parameter to cause IOS to send RADIUS requests to the server on that port.Advanced Security Services client does not support auditing of command or system events on the router or network access server.2. 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.6. 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. IOS always uses 1645 unless you specify otherwise.4. where different sets of security servers are used for different groups of users or different purposes. Central(config)# ip radius source-interface loopback0 Central(config)# radius-server host 14.18 auth-port 1812 Under IOS 12. ! RADIUS example .a group with one server in it Central(config)# aaa server group radius radGroup1 Central(config-sg)# server 14.1. For a complete list of RADIUS router configuration commands see the “RADIUS Commands” section in [1]. you can define named groups of RADIUS servers.1. East(config)# radius-server host 14. Some RADIUS servers use the old standard port 1645 for authentication.18 auth-port 1812 Central(config-sg)# server 14. Also. The example below shows how to set up RADIUS on the router Central.1 or later.2. These groups may be useful for large enterprises. as shown below. To perform authentication and authorization a RADIUS server uses attributes.18 Central(config)# radius-server key W@t7a8y-2m@K3aKy RADIUS servers are freely available and are in extensive use. To define a server group.6. while others use the new standard port of 1812. For more details see the Security Configuration Guide on "Configuring RADIUS" and "RADIUS Attributes" sections for more details. the RADIUS service should be bound to the loopback0 interface.6.4. The shared secret key should be at least 16 characters long and follow the other rules for a good password as described in Section 4.1c 193 .18 key i*Ma5in@u9p#s5wD Central(config-sg)# end Central# Version 1.

Simple example for Central: Central(config)# tacacs-server host 14.4. TACACS+ uses attribute-value pairs for controlling authentication and authorization services.1c . 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. For more details on the TACACS+ and attribute-value pairs see the Security Configuration Guide sections “Configuring TACACS+” and “TACACS+ Attribute-Value Pairs”. 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. For a complete list of TACACS+ router configuration commands see the "TACACS.2. IP address or DNS name. • tacacs-server key string command sets the default TACACS+ shared encryption key. 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. name it in a method list instead of the default group ‘radius’. 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.6. 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 [port integer] can be used to specify a new port number. and users never have to type it. Extended TACACS. you can define named groups of TACACS+ servers. The security of TACACS+ depends on this secret. so make it longer than a typical login password.1 or later. These groups may be useful for large enterprises.1. where the TACACS+ server is running. and TACACS+ Commands" section in the "Security Command Reference". TACACS+ is implemented by Cisco using the AAA mechanisms and provides for the centralized validation of users using routers and network services.Router Security Configuration Guide To use a server group. TACACS+ protects communications using a shared secret key between the network device and central server. Under IOS 12. where different sets of security servers are 194 Version 1.18 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. These attribute-value pairs are configured on the server and used by the router authorization mechanism to control access to network services.

18 key Gx98-vAR1bv*u Central(config-sg)# server 14. These two topics are outside the scope of the Kerberos coverage in this guide. In this system a client would request the credentials of the party they wish to contact from the trusted authentication service. person or network component) on unprotected networks. After the two parties have been authenticated (in this case. Kerberos can also be used to perform EXEC shell authorization using Kerberos Instance Mapping. This mechanism can verify the identities of two users (i. but more details can be found in the IOS documentation [1]. then this form of centralized authentication may be a way to gain excellent security for remote administration. make sure that the IOS installed on your router supports Version 1. consult [8]. simply use the group name instead of the default name ‘radius’ or ‘tacacs+’. Kerberos can provide very effective confidentiality and data integrity services.a group with two servers in it Central(config)# aaa server group tacacs+ myTacGroup Central(config-sg)# server 14.10. Some modern operating systems come with Kerberos built in. If you already have a Kerberos infrastructure in place. as shown below. To define a server group. as well as in RFC 1510 [5] and in Tung’s book [8]. Host configuration for using MIT Kerberos is not covered in this guide. Kerberos provides strong authentication for client/server applications by using secret-key cryptography. consult the IOS 12.Advanced Security Services used for different groups of users or different purposes. Configuration of Kerberos installations based on MIT Kerberos are already explained in the Cisco IOS documentation. ! TACACS+ Example .2. This authentication is performed using a trusted third-party service using conventional (secret key) cryptography. Before attempting any of the step below.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 1510.39 key t777+08cdcoWW Central(config-sg)# end Central# When you want to include the servers of a particular group in a method list. The communications between the router and the Kerberos security server are encrypted. the router and the adminstrator). There are free open sources versions of Kerberos available as well as commercially supported products. use the command aaa server group. This section assumes basic familiarity with Kerberos administration and security concepts. Note that Kerberos only allows for limited authorization capabilities and no accounting.1 Security Configuration Guide for more information. Configuration of a Microsoft Windows 2000 Server acting as the Kerberos authentication server is covered below. For a good introduction to these topics.6.2. if your Telnet client supports Kerberos encryption.e. Kerberos infrastructures are already in wide use.

(Note: this is a Kerberos identity for the router. and select “New”. (For example. right click on the “Users” folder. To make it work with a Cisco router. not for any user. perform the following steps on your Windows server: 1.Router Security Configuration Guide Kerberos. Update or confirm the DNS entries for the KDC and the router. then “user”. 3. 2. create the user accounts on the server for administrators that will access the router. The tools are found in “support\tools\setup.exe”. if you get several choices then your IOS supports Kerberos. Install the Kerberos support tools from the Windows 2000 installation media. Create a user account for the router.) 4.1c . in global config mode type the word kerberos and then type a question mark. 196 Version 1. If necessary.) A Windows 2000 Server configured to be a Domain Controller automatically has the Kerberos Key Distribution Center services installed and running on it. Open up the “Active Directory Users and Computers” tool located in the “Control Panel\Administrative Tools” folder.

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. Includes some coverage of access lists and passwords. and Perkins. 1st edition. 2000. Addison-Wesley. 1998.. Cisco Systems. “Increasing Security on IP Networks” Cisco Internetworking Case Studies. Cisco Security Advisories. and Smith.htm Initial documentation on unicast reverse-path forwarding verification. “Secure IOS Template . This detailed guide explains a great deal about operational use of Cisco routers in the Internet Service Provider environment. E. available at: http://www. Routing in the Internet. McGinnis.2.1”.1(CC) Release Notes.3 or later. protocols. April An old but useful article on using a Cisco router to protect a network boundary.M. 2nd Edition. 2000.Advanced Security Services Coulibaly.. 3. Cisco IOS Release: The Complete Reference. features in releases. Cisco Press. and upgrade paths. available at: http://www. including both standard and vendor-specific structures. the release management process. Articles and Papers Thomas.1c 203 . it includes a good explanation of the concepts.. 2002. P. 1996. It has also been published as a book. including good coverage of critical security topics. November 2003. available from Cisco Press.7. Huitema. “Unicast Reverse Path Forwarding”.com/warp/public/707/21. A deep and detailed textbook about IP routing technologies. 4. Unbelievably detailed information on Cisco IOS release versions. available at: http://www. Version 1. and how routing works in the Understanding SNMP MIBs. available at: http://www. A detailed exploration of the SNMP management information base. “Improving Security on Cisco Routers”. Cisco Press. M. software/ios111/cc111/uni_rpf.html This short but highly prescriptive document distills a great deal of Cisco IOS security practice into an example configuration. Cisco ISP Essentials.html A good overview article on tightening up the security on a typical Cisco router running IOS 11. Greene. B..

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

we assume the routers have two modes of operation: basic mode and privileged EXEC mode.g.12. will be used to help demonstrate the concepts (see Figure 4-1). IP Network Security Prior to establishing an IPSec configuration on the router.20 eq 500 Also. The security guidance of this section does not exhaustively cover all IPSec options.0.1. a pre-shared key) to do it. and the User Datagram Protocol (UDP) port 500 in its communications.1. and the Remote router.20.1.2. host 14. In the basic mode of operation. certain network and current router configuration checks should be made to eliminate any router connectivity problems. Since IPSec utilizes IP protocols 50 and 51.20 access-list 100 permit udp host 7.e.12. Establishing a common IKE Authentication Key 2.20.2. 7. it provides a set of options (e. access-list 100 permit 50 host 7.12.20 host 14.1.20 host 14. Version 1. the external interfaces of the North router. which algorithms to use) and the appropriate Cisco IOS commands to implement them in an easy-to-follow. see Section 4. While IKE has several ways it can authenticate the two routers to each other. Establishing the IPSec Protection Parameters Establishing a Common IKE Authentication Key Prior to establishing an IPSec tunnel between two routers. Building IPSec Tunnels Building IPSec tunnels between two Cisco routers will involve entering three sets of information into each router’s running configuration files. we will only discuss how it uses a jointly held secret value (i. the administrator can update and/or change the current running configuration. each router authenticates the peer IP address with which they are building a tunnel.2. 14. 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. anyone with access to the router can view selected information about the current running configuration. for operational security we HIGHLY recommend that IKE authentication decisions be made using IPSec authentication schemes in conjunction with digital certificates.12.0. This authentication decision is made in the IPSec framework using the IKE protocol. The example below illustrates the ACL rule syntax for permitting incoming IPSec traffic. However. For more information about command modes. Establishing an IKE Security Policy 3.2.1c 205 . For the example in this section. In the privileged EXEC mode.Advanced Security Services 5.0.20 access-list 100 permit 51 host 7. step-by-step example to help you set up and test IPSec on your network. The sets can be labeled as: 1. In the example that follows.2. Rather.0. the routers may be configured using several different modes of operation.2.

to check and see if the router has received the intended configuration information. North(config)# crypto isakmp key 01234abcde address 7. Do not use a remote administration connection to enter sensitive IPSec parameters – use a local console connection. each router must possess the same secret key. This can be done using the show running-config command in privileged EXEC mode.2. These keys should be obtained out-of-band by each of the routers’ administrators. one per line. The syntax for the crypto isakmp command is: crypto isakmp key key-value address destination-ip-address. there must be at least one policy in common between them. We strongly recommend using difficult-to-guess combinations of characters. End with CNTL/Z. Once the keys are securely held.) To use pre-shared keys for making authentication decisions in IKE.Router Security Configuration Guide Consult the Cisco IOS 12.0. the secret key is “01234abcde”. End with CNTL/Z. and that name appears in all the prompts.0 Security Configuration Guide [2] for details on the other IKE options. and punctuation symbols to build operational pre-shared keys. In order for two routers to be interoperable. Each necessary parameter for the policy is listed 206 Version 1.1c .1. the network administrators for the North and Remote routers (possibly the same person) should enter the key into their routers. as shown below.20 Remote(config)# exit Remote# When entering new configuration information into the router it is always a good idea. (Note: the router used for part of this example is named “Remote”. North# config t Enter configuration commands. To enter the keys. use the crypto isakmp command in global configuration mode. These policies capture information needed by the IKE protocol to help build a secure IPSec tunnel between the two routers. one per line. For this example. Remote(config)# crypto isakmp key 01234abcde address 14.12. after entering the new information.20 North(config)# exit North# and Remote# config t Enter configuration commands. numbers. Establishing an IKE Security Policy Each router contains a list of IKE security polices. 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.

one per line.g. however. the Internet). ! 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. 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. 12. pre-shared keys) • Diffie-Hellman group – used for computing the encryption key (choices: #1 (768 bit modulus). (e. #5 (1536 bit modulus)).536. The DES algorithm is not acceptable. and community of interest data separation is needed. North# North# config t Enter configuration commands. group #5 should be used where possible. to protect information between two peers over a hostile. RSA encryption. so use 3DES for such cases. North(config)# crypto isakmp policy 1 ! The policy number may be an integer between 1 and 65. Unless you have a very sound reason to use DES. #2 (1024 bit modulus). e.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.1c 207 .0(3)T). 3DES doesn’t provide the needed performance) always use 3DES. 3DES in certain IOS versions. unprotected network (e. otherwise use group #2 • security association lifetime – lifetime.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. MD5) • authentication method – for identifying the routers attempting to establish a tunnel (choices: Rivest-Shamir-Adelman (RSA) signature. expressed in seconds or in kilobytes transferred. End with CNTL/Z.g. • hash algorithm – for providing integrity to IKE protocol messages (choices: SHA.

g. IKE will determine preliminary information needed to create IPSec tunnels. ftp. and ! community of interest data separation is needed.Router Security Configuration Guide Remote# Remote# config t Enter configuration commands. all telnet. with ! the priority given to lower numbers Remote(crypto-isakmp)# encryption 3des ! If the user’s version of the IOS only supports DES.536. We now need to give the tunnel its desired characteristics. End with CNTL/Z.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. one per line. no volume limit Default protection suite encryption algorithm: DES . and http traffic). 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. The following example displays an access list needed to protect ALL protocol information between the North and 208 Version 1. Remote(config)# crypto isakmp policy 1 ! The policy number may be an integer between 1 and 65. Creating the appropriate access lists Some administrators will want to create tunnels to protect all protocol data flowing between two routers. This parameter set can be built using the following three steps: 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.1c . 0.0.0. If the tunnel needs data authentication protection. North(config)# access-list 161 permit ip 7. Choosing the any option for the source and destination also eliminates the need for netmasking in the access list.255 Remote(config)# 2.255 14.255. End with CNTL/Z. 0.255 14.255 North(config)# access-list 161 permit ip 0.255 North(config)# and Remote# config t Enter configuration commands.255.255 14.0. is shown below. North(config)# access-list 161 permit ip 0.0.255 Remote(config)# access-list 161 permit ip 14. see Section 4.0 0.255.255. one per line.255 7. 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. Configure the appropriate transform set The Cisco transform set identifies the desired protection mechanisms for building the IPSec tunnel. Access lists can be used to improve the granularity of the IPSec tunnels.0.0.255. access-list 161 below) for both the source and destination in the access list will force all packets to be IPSec protected.255 Remote(config)# access-list 161 permit ip 7.255. 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. then choosing either the Authenticated Header (AH) or the Encapsulated Security Payload (ESP) IPSec protocols with either hashing algorithms SHA or MD5 will suffice. North(config)# access-list 161 permit ip 14.1c 209 . 0.0 0.255.0 0.255.255 7.255 Remote(config)# access-list 161 permit ip 14. If the tunnel you are setting up needs data confidentiality protection. End with CNTL/Z.2. Remote(config)# access-list 161 permit ip 7.0 Using the any option (e.0.0.2. then choose the ESP protocol with either the DES or 3DES encryption algorithms (we highly suggest 3DES).0. one per line.Advanced Security Services Remote routers.3 to learn more about access lists.255 7.0 0.0 0.255. somewhat simplified.0 0.0 0. The syntax for an access list rule.g. 0.

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

The following example shows the construction of a single crypto map for the North and Remote routers. See “Configuring IPSec Network Security” in the Cisco IOS 12. The syntax for the crypto map command is: crypto map map-name seq-num ipsec-isakmp.0 Security Configuration Guide to learn more about crypto maps.1. Configure the IPSec crypto maps using the following commands: North# North# config t Enter configuration commands. while the sequence numbers are used to set the priority for two or more crypto maps with the same name. End with CNTL/Z. If two or more crypto maps with the same name are used. Each crypto map is identified by a map-name and a positive integer sequence number (called seq-num below). End with CNTL/Z. 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. which combine the previously entered configuration information. one per line. 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.0. those with lower the sequence numbers have higher priority. 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 1.2.Advanced Security Services whether to use the IKE protocol in setting up the tunnel. The map-name used can represent one or more crypto maps.1c 211 .12. 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 14.20 North(config-crypto-map)# set transform-set set1 ! The following are optional.20 Remote(config-crypto-map)# set transform-set set1 ! The following are optional. one per line.

End with CNTL/Z. North# config t Enter configuration commands. 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.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. 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. then remove the crypto maps from the interfaces which they were applied. North# config t Enter configuration commands. As a quick check (which could potentially eliminate many headaches) before turning on IPSec.1c .12. 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. If both routers are connected to the Internet. North(config)# interface ethernet 0/0 North(config-if)# no crypto map pipe-1 North(config-if)# end North# and 212 Version 1. Assuming the ping was successful. without an IPSec tunnel).12. using outside interface eth0/0. one per line. End with CNTL/Z.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. make sure the two routers are in a state where they can communicate (i. we are now ready to build a tunnel between our routers. as in Figure 4-1. in all likelihood.e. give us this answer.20 on North should. we are almost ready to build a tunnel between the North and Remote routers.1. one per line. one per line.1. as shown below. End with CNTL/Z. A simple ping 7.

etc.2.20 remote crypto endpt. 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 path mtu 1500. local addr.e. 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.12. flags={origin_is_acl.255.20 14.12.20/255. #pkts decrypt: 5. #pkts verify 5 #send errors QM_IDLE conn-id 1 slot 0 North# show crypto ipsec sa interface: Ethernet0 Crypto map tag: pipe-1.20 local ident (addr/mask/prot/port): (14.: 14. #pkts encrypt: 5.0. the user should be able to review the IPSec tunnel parameters. #pkts digest 5 #pkts decaps: 5. End with CNTL/Z. 14.255.12. These parameters can be seen using the show crypto ipsec security-association and the show crypto isakmp securityassociation commands.: 7. 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.1.255.} #pkts encaps: 5.2. one per line. Depending on the time allotted for a ping echo reply to return to the ping source. North# show crypto isakmp sa dst src state 7.Advanced Security Services Remote# config t Enter configuration commands. If the negotiation is successful. the tunnel will be established and the ping requests will be protected.2.20/255. media mtu 1500 current outbound spi: 1B908AE inbound esp sas: spi: 0xEFA038E(251265934) Version 1. the access lists will have notified the IOS that an IPSec tunnel has been requested to protect packet data.0.1.20 PERMIT. the transform set). Once the IPSec tunnel has been established.12.1.255/0/0) remote ident (addr/mask/prot/port): (7.0.1. If everything has been set up properly.1c 213 . speed of the network.0.255/0/0) current_peer: 17. #recv errors 0 local crypto endpt.

in use settings ={Tunnel.Router Security Configuration Guide transform: esp-3des esp-sha-hmac . the following suggestions will help reset the IPSec relevant router parameters and hopefully allow for a tunnel to be constructed. can be very temperamental.1c . Hence. } slot: 0. To remove this information and allow the routers to begin a fresh IPSec negotiation of tunnel parameters. conn id: 2. First. conn id: 3. 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. 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 . the construction of the IPSec tunnel will not succeed. And even when a tunnel is established. if the crypto maps are removed from the interfaces where they were placed (e. 214 Version 1. in use settings ={Tunnel. this information will reside in IOS memory and hinder future attempts at constructing tunnels between these two peers.g. then the information will be removed. all tailored to the specific peer devices involved. The EXEC mode commands clear crypto sa or clear crypto isa commands. several things can be done. several commands may be used to collectively remove the unwanted information. and the global configuration mode command no crypto ipsec sa. including Cisco’s. 1. If the tunnel fails to be constructed. } slot: 0. will remove the unwanted information. then removing them is not a viable option. 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. If the crypto maps are in use by established tunnels. 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 outbound ah sas: Troubleshooting Most current IPSec implementations. the IOS stores certain information about both of their IPSec configuration files. If any one of many parameters are not set properly. If your routers do not correctly establish the IPSec tunnels that you need.

20 (I) QM_IDLE Version 1.0. turning on full debugging using the debug all command supplies even more diagnostic information.20/16). On North: access-list 101 permit ip 14.0. and determine exactly where the negotiations are failing. Full debugging imposes too great a load to be practical for operational routers.0.0.1. 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.Advanced Security Services 2. 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. timeout is 2 seconds: .0. 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.20 0.0.0.!!!! Success rate is 80 percent (4/5).) 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.1.255 7. Make sure the routers have mirror access lists The Cisco IOS IPSec code can get easily confused when the access lists.12.2.12. but both were on while the IPSec tunnel was being constructed.20.255.20 0.1. are not mirror images of each other. On the other hand.0. if we only want to protect packets to/from a LAN behind the Remote router (IP address 7.0.1 0.29 (I) QM_IDLE 00:19:35: ISAKMP (1): received packet from 7.2. Sending 5.0. (Note: These debug options were run at different times.1c 215 . In our example above. 100-byte ICMP Echos to 7. which are engaged by the crypto maps to determine what packets are protected using the IPSec tunnel. with source and destination addresses each behind one of the routers.1 0. M-ID of 405257172 00:19:35: ISAKMP (1): sending packet to 7.255.12. which can be entered while the router is in privileged EXEC mode.255 3.2.20 Type escape sequence to abort.0.12.1.) North# debug crypto isakmp Crypto ISAKMP debugging is on North# ping 7.1.1. (Note: If the routers establishing the IPSec tunnel are not currently operational.255 14. round-trip min/avg/max = 32/33/36 ms North# 00:19:35: ISAKMP (1): beginning Quick Mode exchange. get protected.255 On Remote: access-list 102 permit ip 7. Below is a list of the North router’s output when these two debug commands were turned on.1.1/24) with anyone behind the East router (IP address 14.0.

(key eng.1. spi= 0x0(0). conn_id= 0. transform= 3esp-des esp-sha-hmac . message ID = ISAKMP (1): Checking IPSec proposal 1 ISAKMP: transform 1.2. msg.20 (I) QM_IDLE North# no debug all North# debug crypto ipsec Crypto IPSEC debugging is on North# ping to 14. flags= 0x4 4w0d: IPSEC(key_engine): got a queue event.12. 4w0d: IPSEC(initialize_sas): . message ID = ISAKMP (1): Creating IPSec SAs inbound SA from 7. dest_proxy= 7. protocol= ESP.255/0/0 (type=1).1.20 (proxy 7.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. dest_proxy= 7.12.20/ transform= 3esp-des esp-sha-hmac .20 ) has spi 59056543 and conn_id 4 and flags 4 lifetime of 3600 seconds lifetime of 4608000 kilobytes outbound SA from North# 4w0d: IPSEC(validate_proposal_request): proposal part #1. msg.1. src= src= 14.20 to 7..1.20 to 14.0.0. src_proxy= 14.20/255. 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.20 ) has spi 595658916 and conn_id 5 and flags 4 lifetime of 3600 seconds lifetime of 4608000 kilobytes ISAKMP (1): sending packet to 7.1.20/255.20.) dest= 7.255.20. src_proxy= 14.255/0/0 (type=1). 4w0d: IPSEC(spi_response): getting spi 595658916 for SA from . protocol= ESP.20.255.20 for prot 3 4w0d: IPSEC(key_engine): got a queue event.255.) dest= 7.12. lifedur= 0s and 0kb. (key eng.2. 216 Version 1. lifedur= 3600s and 4608000kb.1.0. keysize= 0.20.20 to 7..1.1. message ID = ISAKMP (1): processing ID payload.1.255.20 (proxy 14.0.255/0/0 (type=1).12.2. ISAKMP (1): processing NONCE payload.255/0/0 (type=1).20 to 7..

0.9. First. perform the following steps: 1. like that obtained from running the debug commands on the router. or isolating networks within an organization. eliminating the threat of a network sniffer seeing passwords or sensitive configuration information.0. we’ll set up the configuration on the router. Configuring a Cisco Router for IPSec Secure Remote Administration On the Cisco router. The IP address of the North router on the interface closest to the administrator is 14.255.2. sa_prot= 50.20.20/255.1. and for recovering from failures. keysize=0. transform= 3esp-des esp-sha-hmac . In this example.255. sa_conn_id= 101 North# no debug all 4. sa_trans= 3esp-des esp-sha-hmac .) dest= 7. so we’ll secure a connection to there. Use an IP packet sniffer to observe the contents of each packet in the IPSec tunnel negotiation This information.1c 217 .2. dest_proxy= (type=1). (key eng. sa_prot= 50.2. lifedur= 3600s and 4608000kb. sa_trans= 3esp-des esp-sha-hmac . sa_spi= 0x238108A4(595658916). is invaluable in diagnosing exactly where the tunnel negotiation is failing. a computer on the desk of the administrator is to be used to administer the North router. Common ways to perform administration of a Cisco router are to use either telnet (which sends the password in the clear) or SNMP. msg. conn_id=100. which is next to the servers in Figure 41.1.255.flags=0x4 4w0d: IPSEC(initialize_sas): . sa_conn_id= 100 4w0d: IPSEC(create_sa): sa created. Since both of these run over IP.1. (sa) sa_dest= 7. conn_id=101. Another use of IPSec would be to use it to protect the administration of a Cisco router. (sa) sa_dest= 7.1.255. 5.20. keysize=0.20/ sa_spi= 0x385219F(59056543).Advanced Security Services spi= 0x238108A4(595658916). then examine the configuration sequence for a PC running Microsoft Windows 2000 or XP.6.12. IPSec can be used to encrypt this communication.255/0/0 (type=1).12. 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). This could represent either connecting widely separated networks. src= 14. flags=0x4 4w0d: IPSEC(create_sa): sa created. Let’s say the computer the administrator uses to configure the router has IP address 14.2.250. src_proxy= 14. spi= 0x385219F(59056543). Enter configuration mode: Version 1. protocol= ESP.2.20.

Pre-shared keys are passwords – or better yet a passphrase. and SHA. and so pre-shared keys can be used. numbers. Enter the authentication password.e. Enable telnet access to the router for administration from the administrator’s machine. However. 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. The other option is the lifetime until a key renegotiation is required. SHA. The policy number selected here is 10. or anything easily guessed. if some form of a public key infrastructure (PKI) is already in place. but again. include letters. it is not forwarding this particular traffic anywhere else). The key exchange size of group 2 is larger than that for group 1. if two or more ISAKMP policies exist on North. The default hashing algorithm. The transform-set contains the parameters for protecting the actual traffic. so we will skip this. Again. this does not concern us too much. so again we select the stronger option. so we recommend 3DES.1c . 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 1.2. certificates can be used.6 North(config)# line vty 0 4 North(config-line) access-class 12 in North(config-line) exit North(config)# 3. certificates just for this probably aren’t warranted. We’ll use access list 12 to list the machines that may to telnet to the router. which is just an arbitrary number to set a priority.2. we can use transport mode instead of tunnel mode. North(config)# no access-list 12 North(config)# access-list 12 permit 14. so we will take the default and not enter it. we want to use 3DES.9. This choice is also made because currently it is easier to configure IPSec in Windows 2000 to use transport mode.Router Security Configuration Guide North# config t Enter configuration commands. The encryption options are DES and 3DES. DES has been demonstrated to be weak. one per line. Create an ISAKMP policy. North(config)# 2.5 for more guidelines on password quality).1. North(config)# crypto isakmp key my4pa$$phra$eHere address 14. Since the authentication is only between 2 machines. End with CNTL/Z.6 5. and possibly not even strong enough to protect passwords. Since we are treating the router as just a host to connect to (i. is suitable. Please do not use anything in the dictionary.9. and punctuation (see Section 4.

9.1. North(config)# access-list 167 permit ip host 14.9.6 host 14. and IPSec should be running on the Cisco router. Version 1. Then we exit from configuration mode. “Close” the list of additional snap-ins. run Microsoft Management Console (MMC). we’ll ensure that no other crypto maps are still in existence before we define this one. The management console window should now look something like the screenshot below. (The screenshots shown below were captured on Windows 2000.2. or by using the “Run” command from the “Start” menu).250 host 14. The match address links the desired access-lists to the crypto map. Finally. A crypto map must be created. The IPSec connections must be allowed. so we use the one we entered in the previous step. 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. Any name can be given to this – we use ciscoadmin.1. You can add the snap-in by looking under the “Console” menu and selecting “Add/Remove Snap-in”.9. “Add” the IPSec snap-in for the local machine. but the directions also apply to Windows XP. initially empty. Scroll down until you see one titled “IPSec Policy Management” or “IP Security Policy Management” and select that one. the Windows 2000 computer on the desktop of the administrator can be prepared. Priority for this crypto map is set to 10.250 interface is named Ethernet 0/0).6 set transform-set 3des-sha-xport match address 167 exit 8.1c 219 . When doing so.1. and select “Local Computer”.2. 167. A few of the dialogs may look slightly different. and “OK” the one snap-in that you have added.6 North(config)# access-list 167 permit ip host 14. Click the “Add” button and you will see all the possible snap-ins.1.250 7.1. either from a command window prompt. It will ask which computer it should manage. We number the access list as 167. That will give you a window containing the currently added list of snap-ins. 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. click “Finish”.) First.Advanced Security Services 6.2. apply these definitions to the interface (the 14. This section assume moderate familiarity with Windows 2000 or XP network administration.2.

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

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

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

click “Add” and you should see the following screen. Version 1. This third wizard is not helpful.Advanced Security Services We enter that in the appropriate box. From there. unselect the “Use Add Wizard” option. click “Add” and the following IP Filter List definition window will appear. So. but before you “Add”. 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. If you use the wizard. Name this filter (Cisco Only Filter. The IP Filter List window will appear. or something like that). and click “Next”. it is probably empty.1c 223 . Initially. Now we need to add a filter. unselect the “Use Add Wizard”.

1c . Double click on “Require Security” to see what options are set. 224 Version 1.255. “Permit”.255 which permits secure connections only to the one router and leaves all other communications unaffected. Use a subnet address of 255. Click OK. you will want to examine it in a bit more detail to be sure that it contains the options you need. returning you to the filter list window.” Then select that filter (call it “Cisco Router Filter”) from the list of filters and click “Next”. 14.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. “Request Security” and “Require Security”. Before selecting the “Require Security” option. It should look something like this. The next window that appears is the Filter Action window. You do need to leave the mirrored option on so filters are defined for traffic going in both directions. There are three default filters defined.1.2.255. which you should “Close.250.

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

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

1. North(config)# 2. make sure that the router has a local hostname and domain name set. IOS 12.1c 227 . There are two main versions of the SSH protocol in widespread use. see Section 4.6 for detailed information on how to define usernames. Because the connection is encrypted. The detailed example below shows how to configure the router North (14.2. one per line.3. For more information about access lists. 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. For more information about IOS versions. you must establish usernames for people that will be attempting to connect to the router.1. First. Configuring a Router for Secure Remote Administration with SSH While in IOS enable mode. North(config)# no access-list 12 North(config)# access-list 12 permit host 14. and rcp services.Advanced Security Services 5. SSH versions 1 and 2.3. For a thorough discussion of SSH. rlogin. there are two important prerequisites to address. It utilizes RSA public key cryptography to establish a secure connection between a client and a server.9.1 to connect to the router for administrative purposes. With SSH. In this scenario the administrator will use an SSH client on the host 14. See Section are currently capable of supporting only SSH version 1. Enter configuration mode: North# config t Enter configuration commands. 5.2. Instructions for this can be found in section 4.2. consult [13].4 can support SSH version 2. Before you can configure SSH. SSH also prevents session hijacking and many other kinds of network attacks. Versions after and including 12. commonly called SSH. passwords and other sensitive information are not exposed in the clear between the administrator’s host and the router.2. see Section 8. This example uses standard IP access list 12 to identify the hosts that may start SSH sessions into router North. Cisco IOS 12. Configure an access list permitting access from the administrative host. perform the following steps: 1.0(5)S with IPSec include support for SSH.9. End with CNTL/Z.0 through 12.1 log North(config)# line vty 0 4 North(config-line)# access-class 12 in North(config-line)# exit Version 1.3.250) to run the SSH server and accept incoming connections. IOS versions that do not support IPSec usually do not support SSH either. SSH was originally intended to be a secure replacement for classic telnet. Only certain Cisco IOS versions are shipped with the SSH feature set. rsh.

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

which is a sound choice. and verify your output states that SSH is enabled.Advanced Security Services • The number of incorrect login attempts that are permitted before the router will drop a remote access connection is also configurable.5 Authentication timeout: 90 secs. North(config)# ip ssh authentication-retries 2 North(config)# To disable telnet and require SSH.3. The default value is 3 attempts.2 To verify that SSH has been successfully enabled. these commands vary between IOS releases. It is also possible to accept both SSH and Telnet. execute the following command. execute the following command. which is recommended.5 North# Encryption State 3DES Session Started Username joeadmin IOS Version 12. the maximum value is 5. If your session is secure then the output should resemble that shown below.1c 229 . North(config)# line vty 0 4 North(config-line)# transport input ssh North(config-line)# exit North(config)# 5. North# show ip ssh SSH Enabled – version 1. North# show ssh Connection Version 0 1. simply leave off use the transport input command as shown below.2. There may or may not be any current connections. and verify your output is similar to the following. Authentication retries: 2 North# To verify that SSH has been successfully enabled and check that your session is actually using SSH. The two subsections below describe the commands in detail. depending on how you are connected to your router. the example below shows how to set the router to drop the connection at the second failure. Do not set the value higher than three. connect to the router using your SSH client and type the command show ssh. North# show ip ssh Version 1.1 To verify that SSH has been successfully enabled. IOS Version 12. Advanced SSH Commands There are a number of commands that will allow you to verify that the SSH server is now operational. but if an IOS router supports SSH then Telnet should not be used. However.

1. or that a connection must be forcefully closed.5 Authentication timeout: 90 secs. 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.1. which is in use by user joeadmin.5 Encryption 3DES State 4 Username joeadmin Diagnosing and Managing SSH In the unlikely event that a connection has not properly closed. % 230 Version 1. % ssh –l joeadmin 14.1. Obtain the connection number using one of the two methods shown above. Authentication retries: 2 North# show ssh Connection Version Encryption State Username 1 1.1.250 joeadmin@14.1c .5 3DES Session started joeadmin North# exit Connection to 14. there are two ways to go about this.2.250 closed. then use of the commands below to disconnect the session. It is very important to disable debug messages when you are finished using them.1.Router Security Configuration Guide Connection 0 North# Version 1. North# disconnect ssh 0 North# .or North# clear line vty 0 North# You can use the IOS command debug ip ssh to diagnose SSH operation. This example disconnects session 0.250’s password: Warning: Remote host denied X11 forwarding North> enable Password: North# show ip ssh SSH Enabled – version 1.

There are usable clients available for almost every host platform (even PDAs). configure the router to use it exclusively as shown below.1 (freeware) 5.3.4. Some Available SSH Clients To employ SSH between an administrative host and the router.0 Authentication timeout: 90 secs. consult [12]. For more information about vulnerabilities in SSH and Cisco IOS. Section 9.Advanced Security Services 5.3.version 2. it is still a much more secure choice for remote administration than unprotected Telnet.3. Authentication retries: 2 North# Version 1. Below are some useful clients for common host operating systems.3. 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. but tools that implement some of them do exist. the host must support an SSH client. North(config)# ip ssh version ? <1-2> Protocol version North(config)# ip ssh version 2 North(config)# exit North# show ip ssh SSH Enabled .1c 231 . Security of SSH Versions There are several known security weaknesses with the SSH version 1 protocol. and most Unix and Linux distributions include an SSH client by default. Attacks that exploit these weaknesses are complex and non-trivial to execute. When your IOS release supports SSH version 2. Even though SSH version 1 may be subject to network man-in-the-middle attacks in some circumstance. Information on where to download these applications can be found in the Tools listing.

1c . consult the Cisco IOS documentation for information about IOS NAT features. The figure below shows the basic structure for a CBAC-based firewall setup. Stateful inspection firewalls do this by inspecting each packet for compliance with the specified security policy. To reach even a moderate level of effectiveness as a firewall.) 5. Because routers connect networks together. by transparently translating all IP addresses and coalescing distinct IP addresses into a single one. Using a Cisco Router as a Firewall This section describes how to use a Cisco router as a modest firewall. a router can hide the structure of the trusted network. Current product catalogs and web pages use the new name. and a firewall between a main network and a restricted enclave. Potential applications for using a Cisco router as a firewall include: a modest Internet firewall. Section 4. Using NAT.4.1. if it is running a version of IOS that has firewall capabilities.” Much of the documentation still uses the old name.4. The Cisco IOS Firewall feature set Content-Based Access Control (CBAC) facility allows a router to act as a rudimentary stateful inspection firewall. As long as all traffic between the trusted and the untrusted network pass through the firewall. This guide does not describe NAT.3 describes access lists in detail. Note that CBAC is intended mainly for border routers. CBAC can provide modest firewall protection for a network without extra hardware. it can effectively enforce a number of network security capabilities. (Note: in mid-2000. provide a rudimentary firewall capability in their routers. The security policy for this setup is to permit users to take advantage of certain network services on the untrusted network.) 5.Router Security Configuration Guide 5. the router configuration must include good access lists. including Cisco. Basic Concepts A network firewall is a network device that connects a protected internal network to some other untrusted. (Another important feature for firewalls is hiding network addresses and structure. external individuals from gaining access to your internal network. a firewall between two different communities of interest. while at the same time allowing authorized users on the trusted network (the ‘inside’) access to services on the untrusted network (the ‘outside’). and that is what we will use below. possibly hostile network.4. it offers you another facility for enforcing security policy at the boundary between different networks. but to offer no such services in the other direction. Cisco renamed the IOS Firewall to “Cisco Secure Integrated Software.2. Cisco IOS provides full support for Network Address Translation (NAT). 232 Version 1. Configured together with good access lists. Configuring Cisco IOS Content Based Access Control The Cisco IOS Firewall feature set is designed to prevent unauthorized. and to block attacks on your network. many router vendors.

Response comes back from the web server. maintain state information. or an outbound list on the inside interface. CBAC inspects the initial TCP packet of the connection.Advanced Security Services Untrusted Network 14.20 (port 80) on the untrusted network. and adds a rule to the inbound access list.2.1. Outbound request Router Inbound response trusted network Host Figure 5-2: CBAC Overview Version 1. CBAC adjust access list 3.64 Trusted Network 14. passes access list.7 Figure 5-1: A Simple Router Firewall CBAC examines not only network layer and transport layer information.2.1.2. Because CBAC works by modifying access lists.6.64 Router 14.9.10. Host initiates a web connection to web server 7. CBAC maintains connection state information for individual connections. and use that information to make access decisions. either an inbound list on the outside interface. permitting data from 7. inside interface inspect outside interface 2.0/24 User Host 14.2.20 port 80. but also examines the application layer protocol information (such as FTP information) to learn about the state of TCP and UDP connections. 3.6. there must be at least one access list in place on the path from the untrusted network to the trusted network.10. 1.10.2. The heart of CBAC is the ability to inspect outgoing IP traffic in real-time. The decisions are enacted when CBAC dynamically adds rules to interface access lists to pass permitted traffic. untrusted network 1. 2. The figure below illustrates this.1c 233 .10.6 User Host 14.

3). such as ICMP. and test it from the untrusted network by running a network scanner (see Section 6). The two examples below show a router without CBAC. 7. The easiest way to check is to execute a CBAC-related command. SMTP. if it does not. Example: FTP. It also includes some special case handling for multi-port application protocols. Test the configuration from a host on the trusted network by running services. Steps in Setting Up a Cisco Router Firewall To set up a simple firewall using CBAC. 234 Version 1. This is the access list that CBAC will be modifying on the fly. Set up an outbound access list on the outside interface. Determine the list of services that users or hosts on the trusted network need from the untrusted network. must be separately permitted by the interface access lists if you need them. Apply the CBAC inspection ruleset to an interface.0(9) with Firewall Feature Set 2.5. follow these steps: 1. POP3. and a router with CBAC. There is no simple.Router Security Configuration Guide Note that CBAC handles only TCP and UDP protocols.323 and FTP. then CBAC is not supported.5). or IPSec. Example: IOS 12. Other IP protocols and services. permitting traffic that the router must process. Set the CBAC global timeouts. values that are too long can leave the trusted network vulnerable. Central. usually the outside interface of a border router. like H.1c . 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. South. These timeout values determine the duration of window of accessibility opened back through the firewall in response to a request from the trusted network. 4. if the command fails. 6. Web (HTTP). 5. 8. Check that the router supports CBAC. RealAudio (RTSP) 3. then install an IOS version that does (see Section 4. Call this list the desired services list. Step 1. but prohibiting other TCP and UDP traffic including the desired services list. direct way to check whether a router has CBAC capability. Create a CBAC inspection ruleset supporting the desired services list. OSPF. Set up an inbound access list on the outside interface.

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

3).255 any eq ftp South(config-ext-nacl)# permit tcp 14. For more information about access lists. in general.2.255 host 14. the access list is applied to the outside interface.10. even those on the desired services list. 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.3 eq smtp South(config-ext-nacl)# permit tcp 14. The source address for each rule in the access list should be a network address or address range valid for the trusted network.2.9.2. this is a safe choice. any eq domain South(config-ext-nacl)# permit tcp 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 14.0.2. The example access list below blocks TCP and UDP traffic effectively.10. permits a modest set of useful ICMP messages.1c .2.255 any eq www South(config-ext-nacl)# permit tcp 14.10. or even both. Set up an Outbound Access List Before CBAC can do its work.0.0.0. This access list must permit the protocols on the desired services list.255 host 14.10.0 0.0. An inbound access list intended for use with a simple CBAC firewall scheme should block all TCP and UDP services.0 0.0 0. there must be an access list applied to traffic from the trusted net to the untrusted net.2.3 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. Also. see Section the destination address can be the catch-all any.0 0. In this example. and permits the RIP routing protocol (see Section 4.0 0.0.255 any eq ntp South(config-ext-nacl)# permit udp 14.10.255 any South(config-ext-nacl)# permit udp 14.0 0. in the outbound direction.0.0.255 any eq telnet South(config-ext-nacl)# permit tcp 14. 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 1.0.Router Security Configuration Guide Step 3.2.0 0. this access list must be an extended IP access list. 0. The example below shows an access list for our desired services list.

Enable the alert and audit-trail features to get additional log messages. and the second rule supports web.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. • The connection is idle for longer than an allotted idle time. Telnet. using the command ip inspect audit-trail. the audit-trail option controls whether use of that protocol causes a log message to be generated. use the command ip inspect name. similarly.1c 237 . The syntax is shown below. beyond those generated by interface access lists. Version 1. (In older versions of CBAC. This rule gets removed when one of the following conditions are satisfied: • The response does not arrive within the allotted timeout time. and POP3 email services.) The example ruleset below supports the example desired service list. it adds a rule to the inbound access list to permit the expected response. 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.” Its first rule supports DNS and NTP. • The connection closes down (TCP only). The name of the ruleset is “fw1. audit trails could only be turned on globally. Create a CBAC Ruleset To create a CBAC ruleset. Adjust the CBAC Global Parameters When CBAC detects a connection attempt by a client on the trusted network. 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.

0 are longer than necessary. . South# config t Enter configuration commands. South# config t Enter configuration commands. a modem) or on a highly congested network. 238 Version 1. The example below shows how to set the global timeout parameters.g. End with CNTL/Z.Router Security Configuration Guide The default timeout and idle times in Cisco IOS 12. one per line. South# End with CNTL/Z.1c . Length of time CBAC continues to manage a TCP session with no activity. synwait-time 15 finwait-time 1 idle-time 1800 idle-time 15 Step 7. 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. these values might need to be increased for a very slow connection (e. Length of time that CBAC continues to manage a UDP ‘session’ with no activity. There are also global CBAC parameters related to half-open TCP session. The example below applies the ruleset from step 5 to the outside (untrusted network) interface. Use the interface configuration command ip inspect name to apply a ruleset. 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 . The table below describes the parameters to change.) 15 seconds Of course. Default 30 seconds 5 seconds Suggested 15 seconds 1 second TCP idle-time UDP idle-time 1 hour 30 seconds 30 minutes (1800 sec. Length of time that CBAC continues to manage a TCP session after it has been closed down by a FIN exchange. Timeout Name Synwait-time Finwait-time Description Length of time CBAC waits for a new TCP session to reach established state. Apply the CBAC Ruleset to the Interface CBAC is not in force until a ruleset has been applied to at least one interface. but these can be left at their default values.

.250). Connected to 14.2.255 any permit udp The desired service list for this firewall is: DNS. and POP3 (to a single host). starting a telnet session from a host on the trusted network to a host on the untrusted network. Test the CBAC Configuration Perform some simple tests from a host on the trusted network. CBAC should be running on the router. This sample is formatted as it would appear in a configuration text file stored on a host for download to the router South. The example below shows a Telnet session from a host on the trusted network (14. save the router configuration to NVRAM at this point with the command copy running startup.255 any eq ntp Version 1.2.250 Trying 14.4. FTP. Telnet. see Section 6. as illustrated in the example below. This is the CENTRAL router. SMTP (to a single host).2. NTP. 0.10.10. Access is limited to authorized administrators only! Username: nziring Password: Central> While the Telnet session is active.0 0. Configuration Sample The configuration command listing below shows the configuration commands for a firewall router with a simple CBAC configuration. For more detailed testing information. $ telnet 14. confirming that CBAC is managing the session.0.Advanced Security Services After this step.1c 239 . If the command gives no output.250:23) tcp SIS_OPEN South# If the CBAC configuration seems to be working.10. check the CBAC session status on the router using the command show ip inspect sessions.10.9.2.. The test shown here has two parts: first. It should show the telnet session. Step 8.0. no access-list 110 ip access-list extended 110 permit icmp HTTP.2.9. Escape character is '^]'. and second. 5.6) to a host on the untrusted network (14.2. then CBAC is not working.2. to see that CBAC is working.189:3175)=>(14. South# show ip inspect sessions Established Sessions Session 6187B230 (14.

Router Security Configuration Guide permit udp 14.2.0 permit tcp 0.0.0 permit tcp eq smtp host deny ip any any exit 0.0.255 any eq domain any eq www any eq ftp any eq telnet host 14.3 eq pop3 no access-list 111 ip access-list extended 111 deny ip permit tcp 0.0.255 0.1c . permit tcp 14.0 permit tcp 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 1.255 0.255

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

South# show ip audit all Event notification through syslog is enabled Event notification through Net Director is disabled . configure and apply the audit rules.2 Security Configuration Guide (in the IOS documentation). it is good practice to confirm that the IDS facility is working. the IDS logging facility. and the second the response from an IOS without IDS support (Central).3 for information on IOS versions. it may be possible for you to upgrade your router to a release that does. Configuring the IOS Intrusion Detection System Only those IOS releases marked “Firewall/IDS” support the IDS features described in this section. and Section 8. One way to initialize the facility is to set a parameter on one of the IDS signatures. Second. initialize the Post Office. drop the packet.Router Security Configuration Guide signature list in order to detect possible attacks or suspicious activity.Initialization You must initialize the IDS facility before configuring it. consult the “Traffic Filtering and Firewalls” section of the Cisco IOS 12.5 for information on loading IOS upgrades. See Section 4.2. 242 Version 1. This section presents only a brief overview of the IDS facility. Step 1 . 5. Central# If your router does not support the Firewall IDS facility. When an attack is detected. for more details. depending on how the IDS was configured. the IDS will log an alarm to the syslog server or a Cisco Netranger Director.5. South# config t Enter configuration commands. follow the three steps outlined below to configure them. After you’ve configured everything. South(config)# ip audit smtp spam 200 End with CNTL/Z. The first example below shows the response to an IDS command from a version of IOS with IDS support (South).1c . make sure that your router supports them by attempting to execute a simple IDS command. The IOS documentation recommends using the command below. one per line. Before attempting to configure the IDS features. Third. First. initialize the IDS facility. South# versus Central# show ip audit all ^ % Invalid input detected at '^' marker. and/or reset a TCP session. . Once you have determined that a particular router supports the IDS facilities.

one per line. 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.g.2. End with CNTL/Z. 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. less than 32MB) then you should lower the value to 50.15 localaddress 14. then configure the router to send the alerts to it.Configure the Post Office When the IDS facility detects a match to one of its intrusion signatures.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. The example below shows how to configure the router South to send alerts to an IDS Director at 14. 3.64 South(config)# Version 1.2. Next. set the size of the event queue for the Post Office. South(config)# ip audit po max-events 50 South(config)# Step 2 . If the router has very little RAM (e.10. ip audit notify nr-director This command simply tells the router to use an IDS Director. directing alerts is the job of the Post Office.5 for information on syslog configuration). If your network supports a Cisco Secure IDS Director.15. and the org-id value should be the same as in command 2.1c 243 . 2. ip audit po local hostid host-id orgid org-id This command sets the host and organization IDs for the Post Office. also sometimes called a “Netranger director”.10. and the org-id must be shared by the Director and all Cisco IDS sensors that send alerts to it. above. which is generally a reasonable setting. The host-id value must be the host ID of the Director. The dir-ip-address should be the IP address of the IDS Director. South# config t Enter configuration commands. Setting up the IDS facility to send alerts to an IDS Director requires three commands: 1. see the IOS documentation for details. Each alert must be logged if it is to be useful.10. while the local-ip-address should be the IP address of the router interface closest to the Director. the host-id must be a unique value for this router. it throws an alert.

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

0 Oct 10 20:01:30.1c 245 .3.644 GMT: %FW-4-ALERT_OFF: calming down. The example below shows how to disable two signatures related to common ICMP packet types. This session was captured during a small TCP port scan.696 GMT: %IDS-4-TCP_SYN_ATTACK_SIG: Sig:3050:Half-Open Syn Flood .1.10. This makes the facility most useful for reducing false positives caused by specific trusted hosts (e. South(config)# interface eth0/0 South(config-if)# description External interface.No connections are active South# Version 1. count (51/500) current 1-min rate: 501 . for this case. similar to the one described in Section 6.0 network.1. the router IDS was configured to send alarms to normal IOS logging rather than to an IDS Director.6 to 14. Note that.2. .2. consult the IOS documentation. Applying the rule outbound will reduce false positives.Advanced Security Services only standard IP access lists may be used. 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. a security audit host used to perform test scans).from 14. because only packets which have been permitted by any in-bound ACLs on other interfaces will be scanned.1. For details.from 14. The example below shows how to apply our IDR1 rule for traffic coming into the 14. . you can only restrict scanning by source IP address. 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 .6 to 14.1.752 GMT: %IDS-4-ICMP_ECHO_SIG: Sig:2004:ICMP Echo Request . Oct 10 20:02:19. It is also possible to disable and restrict particular IDS signatures.2. Applying the rule in-bound will yield more complete scanning. South# Oct 10 20:01:13. because all traffic received on that interface will be scanned.1.0 Oct 10 20:01:15.192 GMT: %FW-4-ALERT_ON: getting aggressive.10. 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.g. in either the in-bound or outbound directions.10.

5. it is not analyzed. Without the Director. both require a human to actively monitor them to provide real time information. Many tools and technique exist to help attackers gain this information (e. Special Note: Because it is performed as part of packet routing. The IOS IDS cannot stand alone as a complete network defense package. nor was it designed for this purpose. This can be most efficiently accomplished by installing the IOS IDS at a border point with the firewall configured. The IDS can report intrusions to a given host's syslog.3. Thus.the syslog writes a textfile. which is critical in cases of intrusion. This will provide simple ID at the edge of a protected network and stop simple attacks.1c . they must have information about it. Recommendations The Cisco IOS Intrusion Detection System does not provide comprehensive intrusion detection as a stand-alone feature. 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. Cisco IOS IDS cannot monitor internal traffic. and/or a Cisco Secure Director. one or more dedicated IDS should be deployed on internal networks to provide more comprehensive coverage and analysis. discussed in Section 6. and possibly help you understand threats to your network more quickly. With common attacks stopped and logged. the router console. and attempted root logins. that is. Keep this in mind when planning intrusion detection for your network. while the console is usually in a secure facility. By default.g. It is best used to supplement more complete intrusion detection packages.Router Security Configuration Guide 5.1) The IOS Firewall IDS facility can help detect and track analysis of your network by remote parties. nmap. such as distributed scans. buffer overflows. 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 . because internal network misuse is not uncommon and potentially as detrimental as an external penetration. you cannot use this facility to detect attacks from one host to another on the same LAN.4. if a packet does not need to be routed. Security Considerations for Using IOS Firewall IDS Before an intruder can successfully penetrate a network. 246 Version 1.

com/default. and Virtual Private Networks. A Technical Guide to IPSec Virtual Private Networks. The master document for IPSec. Cisco Product Support Documentation. D. and and related configuration commands.. Cisco Internetworking Solutions Guides. and Atkinson. [7] “Cisco Secure VPN Client Solutions Guide”. and Zwicky. Microsoft Corporation. S. R. 2004. Cooper.0 Network Security. Authoritative source for in-depth descriptions of security-related IOS facilities. [5] Tiller. “Security Architecture for the Internet Protocol. 2000.B. Cisco Press..1c 247 . 2000. A seminal reference for understanding firewalls and the principles for building them. References [1] Chapman.” RFC 2401. [6] “Cisco IOS Firewall Configuration Examples and TechNotes”.com/en/US/products/sw/secursw/ ps1018/prod_configuration_examples_list. [3] Doraswamy. Microsoft Knowledge Base. [8] “How to Configure IPSec Tunneling in Windows 2000”. J. IPSec: The New Security Standard for the Internet. CBAC.en-us. available at: http://www. Cisco Systems.252735 Contains detailed information about setting up IPSec in Windows 2000. plus and technical detail about IKE and VPN design. 2001. available at http://www. N. [2] Cisco IOS 12. This highly technical book provides detailed explanations and pragmatic advice about IPSec.D. Prentice-Hall. IN. 1998.aspx?scid=kb. O’Reilly Associates. D. Intranets.Advanced Security Services 5.html Detailed information on configuring Cisco VPN client software.. Article 2004.6. available at: http://support. [4] Kent. Cisco Systems. includes extensive remarks about VPN architecture.html This page offers several CBAC documents and configuration examples. ps2138/products_maintenance_guide_book09186a008007da16. including IPSec. Version 1. E. 1999. 1999. 252735. 2nd Edition. Auerbach Publications. Contains a good overview of IPSec. Building Internet Firewalls.

Provides broad and detailed coverage of SSH [12] “Secure Shell Version 1 Support”.com/ pcgi-bin/Support/browse/psp_view. 231585. [10] “An Introduction to IP Security (IPSec) Encryption”.cisco. Intrusion Detection.Router Security Configuration Guide [9] “Overview of Secure IP Communication with IPSec in Windows 2000”. Cisco Systems. T. Article software/ios120/120newft/120t/120t5/iosfw2/ios_ids.pdf Provides detailed information on using error messages and debug facilities to track down problems in IPSec configuration. IOS 12. Revision 1. D. and Silverman. A short overview of SSH features in IOS 12. IOS 12. available at http://www. Cisco Systems. [15] “Cisco IOS Firewall Intrusion Detection System”.cisco. available at http://www.231585 A good overview of IPSec features in Windows 2000. Microsoft Corporation.aspx?scid=kb. and usage.1 release 1998. Cisco Systems. Cisco .com/univercd/cc/td/doc/product/ software/ios121/121newft/121t/121t1/sshv1.0(5)T release notes. A good introduction to intrusion detection concepts and techniques. [11] “IP Security Troubleshooting − Understanding and Using debug Commands”. [13] “Cisco Security Advisory: Multiple SSH Vulnerabilities”. Microsoft Knowledge Base. SSH The Secure Shell – The Definitive Guide. 1999. visit: http://www. Wiley. May [16] Escamilla.1c .E. 2001.pdf A detailed overview of the IOS Firewall IDS facility. [14] Barrett. 2000.1(1) An overview of SSH vulnerabilities and IOS versions to which they apply. For more documents and examples about IPSec.pdf This tech note offers detailed information about Cisco IPSec support. with May 2003. available at: http://support. including a list of the supported information and attack signatures. November 2001. software. available at: http://www. 248 Version 1. Cisco Technical Notes. available at: Cisco Systems. Cisco Technical available at: http://www.6. O’Reilly Associates.

Advanced Security Services Version 1.1c 249 .

Routers provide many services that can have severe security implications if improperly configured. Additional tools are listed in the Tools Reference. Attack scripts are readily available on the Internet for numerous well-known exploits.1c .Router Security Configuration Guide 6. and whenever major changes have been made to any part of the configuration of a router. including some versions of IOS. testing should be performed at initial deployment of a router.1.2. this information can then be used to formulate specific attacks against the router. Section 9.3. several denial of service (DOS) attacks and the newer distributed denial of service (DDoS) attacks have been highly successful against network devices. Security testing provides a means of verifying that security functions are compatible with system operations and that they are configured in a secure manner. Packet sniffer programs are used to monitor traffic passing through the network and steal unencrypted passwords and SNMP community strings. 6. Ideally. Principles for Router Security Testing The border router is often the first line of defense when protecting against malicious network attack. Some of these services are enabled by default whereas other services are frequently enabled by users. 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. 250 Version 1.

run a port scan against the router. for IOS v. Otherwise.1.250): Port State Protocol Service Version 1.12 by Fyodor (fyodor@dhp.2. but for IOS v. for TCP FIN use -sF Starting nmap v. the default is disabled.1. use the EXEC mode command show running-config.2.1. there shouldn’t be any ports open.3.2. To view the current operational configuration.2.12 by Fyodor (fyodor@dhp. Testing all allowed services will identify these dependencies.g.1. The above command would not.2. for example.2. The nmap scanning program is a good tool for this purpose. A serious known problem with Cisco IOS is that some default settings are not displayed as part of the router configuration listing. show the ‘udp-small-servers’ or the ‘tcp-smallservers’ in the configuration. For example.1.250 Warning: -sU is now UDP scan. DNS must be available for any operation referencing a host by name to succeed ( Interesting ports on (14.250): # nmap –sT 14. Functional Tests Functional testing provides assurance that the implemented configuration is the intended one. Additionally. 2.250): Port State Protocol Service If VTY (Telnet) access is not allowed. the default is enabled. Telnet).250 –p 1-65535 Starting nmap v. some services depend on other services in order to function. The default settings for these services depend upon the IOS version. cross-check the ports that nmap reports open against the services that the router is supposed to be running. Interesting ports on (14.1. (Note: if IP unreachable messages have been disabled. The examples below show nmap running under Linux. Testing and Security Analysis Techniques 6.Testing and Security Validation 6. as advised in Section 4. UDP Scan: The following command will perform a UDP scan against router North (14.3. To verify the entire configuration.1.11. temporarily re-enable them before performing your UDP port scan by using the interface configuration command ip unreachable. 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.1c 251 .250): # nmap –sU -p 1-65535 14.) TCP Scan: The following command will perform a TCP scan against router North (IP address

252 Version 1. DoS attacks are typically based on high-bandwidth packet floods or other repetitive packet streams.cert. One of the most important security concerns is access to the router itself. telnet to the router. This attack has at least two victims – a target system and one or more reflector systems.1c .org/tech_tips/denial_of_service. The easy availability and effectiveness of DoS scripts on the internet make these attacks a favorite among hackers. Some of the most common attacks on the internet are denial of service (DoS) attacks. Physical security of the router should provide protection from close-in (non-network) access. Once access to the router has been secured. how the router will perform under the stress of an attack. OR EVEN CAUSE THE ROUTER TO CRASH! If the filters are improperly configured. or not applied to the interface. Each packet generates a response from all hosts on the reflector subnet. particularly those without the skill to create their own tools. consult Section 4. The source address in these packets is falsified to be the address of the ultimate target.2. When you do perform testing on the operational network.Router Security Configuration Guide 6. and the Cisco whitepaper “Improving Security on Cisco Routers” [1].. the network is still at risk of attack.3. On the One popular DoS attack is the ‘smurf’ attack. For more information on the effects of DoS attacks. i.html. For a more detailed discussion of Cisco router access security and remote administration. visit the CERT site: http://www. For a general overview of DoS. 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. remote logins should be disabled. visit: http://www. For a detailed discussion of the smurf attack. To test the remote availability.2). including recent developments and links to specific DoS advisories. Attack Tests Attack testing can provide some assessment of the router’s robustness. DO NOT perform attack testing against an operational router without first considering the possible consequences and having a recovery plan. read Craig Huegen’s paper [9]. Connecting to an outside network exposes the internal network and the perimeter router to many potential risks.1. some of these attack tests can have the same effect as a “real” attack from a malicious source. remote access must be limited using authenticated logins or. The router should either refuse the request or prompt for a password.cert. if possible. flooding the target and wasting bandwidth for both victims. Perform testing in a lab or testbed environment before testing in the operational environment. WARNING: RUNNING ATTACK SCRIPTS AGAINST AN OPERATIONAL ROUTER MAY DEGRADE ROUTER PERFORMANCE. The attacker sends a continuous stream of ICMP echo requests (‘pings’) to the broadcast address of a reflector subnet. 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. DO NOT perform attack testing against any network until you have received organizational and legal approval to do so.e.

file transfer or web traffic to legitimate users. Another common DoS attack. Tribal Flood Network 2000 (TFN2K) and Stacheldraht. similar protection can be achieved by filtering for IP spoofing. the SYN flood. The victim responds with a SYN/ACK to the unreachable host and waits for the ACK.7). The server responds with a SYN/ACK packet. the packet is dropped. Additionally. For more options. More information about DDoS attacks is available from references [3]. Also.1c 253 . the ACK doesn’t arrive and the TCP handshake never completes. [5]. An integral part of DoS and DDoS attacks is IP spoofing.3. 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. Blocking access to the service under attack is usually not feasible and would accomplish precisely what the attacker set out to do. Asymmetric routing will not work with this feature. [4]. The effect of this attack is to deny TCP services such as e-mail. In a normal TCP connection request.4. In a SYN flood. these tools employ many compromised systems to simultaneously attack a single target.2 or 11. requiring the attacker to send more phony packets to flood the service. non-routable. packets arriving at the router Version 1. The host could increase the size of the connection queue. arriving packets should not have source addresses of all 0’s or all 1’s or the loopback address (127.0. 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. described below. adds an entry in the connection queue and starts a timer. 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. thus emptying the queue of half-open connections more quickly. However. the timer is reset and the connection is established.0.e. spoofed source address to an open port on the target. Cisco provides a paper titled “Defining Strategies to Protect Against TCP SYN Denial of Service Attacks” [4]. filters can be used to prevent IP spoofing in a manner similar to the ip verify unicast reverse-path feature discussed above.0. The real attacker is able to amplify the DoS flooding while being removed from the attacking machines. four historically popular ones are called Tribal Flood Network (TFN). For Cisco routers running IOS v.Testing and Security Validation Enhanced denial of service tools have recently become available on the Internet. victims of a SYN flood do have some options. the requesting client sends a SYN packet to the server. The host could also decrease the wait time for completion of the three-way handshake.. These distributed denial of service tools (DDoS) pose a major threat to networked systems and have the potential to severely impact normal business activities. takes advantage of the TCP three-way handshake procedure to deny service to the victim. Unlike a “typical” smurf attack. an attacker sends a TCP connection request (SYN) packet with an unreachable. and [8]. 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. This feature checks each packet arriving at the router. and it is only available in IOS v12.0). There are many such tools in circulation. changing the source IP address to hide the true source of the packet. The requester then completes the handshake with an ACK packet. reserved addresses (defined in RFC1918). the queue entry is removed. Trin00.11. which uses a limited number of reflector systems. tracking the attacker is extremely difficult. i.

Some vulnerable IOS devices will “hang” and must be manually restarted by reset or power cycle. At least one commonly available vulnerability scanner can generate these UDP packets. 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. By sending such packets continuously. the access list on the internal interface will recognize such an address as legitimate.0 net access-list 101 deny udp any 14.0. an alternative fix would be to deny all syslog traffic arriving on that interface.) This check will prevent the internal network from being used as a launch point for a source IP spoofing attack. 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.0 eq 514 ! Deny packets addressed to router interface access-list 101 deny udp any host 14.0. 11. If a specific interface is not expected to forward legitimate syslog traffic. as well as at any broadcast or multicast addresses on which the device may be listening. in this case. (Note that the internal network may be using one of the RFC1918 reserved addresses with NAT performed at the router.0. This vulnerability affects unpatched versions of IOS 31. For an in-depth discussion of TCP flooding and IP spoofing. an attacker might be able to completely disable a Cisco IOS device until the affected device is reconfigured to drop the attack traffic.1c .0.20 eq 514 ! Apply to input interface of router North interface eth0 ip access-group 101 in 254 Version 1. This problem can be prevented by applying the appropriate input access list to all interfaces that might receive these packets. The router should drop these packets at the perimeter and log the events.3DB and early (non-GD) releases of 12.0.0 0. The following example shows an access list to block the port 514 UDP traffic.0 eq 514 ! Deny network-specific broadcasts to the 14. To verify outbound anti-spoofing.2. ! Deny all multicasts and all unspecified broadcasts to port 514 access-list 101 deny udp any 224.255.255 eq 514 ! Deny old-style unspecified net broadcasts access-list 101 deny udp any host 0.255 0.2. it might require an administrator to physically visit the attacked device to restore service. RFC 2267 discusses network ingress filtering and defeating DoS attacks which employ IP source address spoofing.2.0 eq 514 access-list 101 deny udp any 14. 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).Router Security Configuration Guide from the internal network should not have a source address that is not one of the legitimate internal addresses. This input access list must block traffic destined for UDP port 514 at any of the Cisco IOS device’s own IP addresses. consult [7]. send a series of packets with modified source addresses to the external interface. the router should again drop the packets and log the events. To verify the anti-spoofing configuration. send packets to the router’s internal interface with source addresses that are not legitimate internal addresses.

always double-check the addresses to be scanned. running DoS attack scripts against the router can have very serious and undesirable consequences. web servers. key applications and routers for those vulnerabilities frequently used by malicious users to investigate and attack networks. after careful consideration. CyberCop offers monthly updates consisting of new modules and security hotfixes for new and evolving vulnerabilities. visit the IIS web site.iss.Testing and Security Validation This vulnerability can be tested by sending a UDP packet to the router’s port 514.pgp. Other sites for exploit information and code are listed at the end of this section. the decision is made to go forward with this testing. Internet Scanner includes nearly 600 total tests. Internet Scanner probes the network’s communication services. Packetstorm Security has several DoS exploits. Version 1. Mechanisms for Automated Testing There are a number of products available to automate the testing process. available under http://packetstormsecurity. security concerns. However. and monitor the tools closely while they are operating. At the time of this writing. or to touch systems for which you have no legal authority. WARNING: RUNNING AUTOMATED ATTACK TOOLS ENTAILS SIGNIFICANT RISK! It is easy to accidentally auto-scan more systems than you intended. the attack scripts are readily available from many sources on the internet. operating systems. For more information.01) now contains tests to find hosts infected by DDoS agents. The latest version of Internet Scanner (6. CyberCop Scanner performs comprehensive evaluations of intranets. For more information. visit: http://www.3. Internet Scanner is also a network vulnerability analysis and risk assessment product. assessment page: the router could crash or hang! As mentioned above.3. if the router is running a vulnerable version of the IOS software and the access list is not properly configured or not applied.1c 255 . level of risk and suggestions for fixing/mitigating the vulnerability. firewalls and screening routers by scanning them and performing extensive tests to identify known vulnerabilities. and http://packetstormsecurity. CyberCop generates reports from scan results that include information about detected vulnerabilities: a description of the vulnerability. Exercise caution when using tools like CyberCop. and updates containing the latest tests and security checks are available for download daily. CyberCop Scanner from Network Associates and Internet Scanner from ISS are two popular commercial products.asp. Internet Scanner analyzes the scan data and provides reports containing vulnerabilities identified along with recommended corrective actions. or SATAN. planning and coordination.php. The Security Administrator’s Integrated Network Tool (SAINT) and the Security Administrator Tool for Analyzing Networks (SATAN) are publicly available tools.

com/.porcupine.2 above. The vulnerabilities discovered are then reported to the user without actually exploiting them. The signature of a smurf attack where router North is the ultimate target would show most of the packets as ICMP echo replies. denial of service attacks are very common on the internet.3. Assume the following access list is applied to interface 14.2. In normal operations. this feature exposes the real security implications inherent in network trust and services. For each problem found. The fourth line therefore matches only packets that are connection requests. 6.Router Security Configuration Guide SAINT gathers information about remote hosts and networks by examining network services such as finger. use the command show accesslist 102. or possibly disabling services. To download a copy of SATAN. In a typical smurf attack. installing vendor hotfixes. The initial data collection can then be used to investigate any potential security problems. For more information. To see the number of matches for each line in the access list. visit ftp://ftp. visit: http://www. SAINT can also be configured to examine trust and dependency relationships in the target network. 256 Version 1. The keyword established in the third line matches any TCP traffic with the ACK bit set. SATAN also provides corrective actions including configuration changes. the attack is probably a smurf attack where North is a reflector. the source addresses in the echo reply packets are limited to a few networks. An analysis of the packets arriving on the serial interface can identify the specific attack being used. a tutorial and the latest version of SAINT.4. SATAN offers a tutorial that explains the problem and the potential impact. any TCP traffic that is not a connection request.3. the TCP SYN packets. ftp. For more information about access lists.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. rsh commands and other services. IOS access lists can be used to characterize the different packet types and to tentatively identify DoS attacks. these are the addresses of the reflector sites. including a FAQ. If the incoming traffic consists mostly of ICMP echo requests. consult Section 4. The third and fourth lines of access list 102 characterize TCP traffic.saintcorporation. Detecting Attacks As mentioned in section 6. NFS.3.0. SATAN was designed to help system administrators responsible for the security posture of their systems.1c . it is a tool for investigating the vulnerabilities of remote systems. SATAN systematically proceeds through a target network probing for common networking-related weaknesses and security problems. a necessary first step in countering DoS attacks. tftp. that is.

3). Also. It is almost impossible to protect a network from denial of service attacks. it may be possible to apply specific filters at the external interface of the border router. Attack Reaction Options It is difficult for the ultimate target of denial of service attacks to stop or even blunt an active attack. First. The best advice is to configure the router to check for IP spoofing.Testing and Security Validation TCP SYN packets account for a third or less of the total TCP traffic. if it can be determined that the originators of the attack are limited to a few addresses. In a distributed attack. if filtering is not possible.3. Second.cert. this filter should only be in place temporarily and only as a stopgap measure. and to only allow services that are needed (see Sections 4. In a SYN flood. SYN floods usually contain packets with invalid source addresses. or not sufficient to stop the attack. An on-going problem is that new attacks can appear so fast on the internet that countermeasures are not immediately available. logging such traffic (as recommended in Section 4. In this case. the only response may be to contact the reflector sites to reconfigure their networks to shut down the attack. 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.5.1c 257 .org) and implementing the latest patches from the vendors. the ultimate target cannot filter out the attacking addresses.3) will let the administrator determine if such source addresses are present. Version 1. both inbound and outbound. This paper gives an overview of denial of service attacks and a detailed discussion of using access lists to categorize packets.2 and 4. The paper also describes how to trace DoS attacks and the complications inherent in packet tracing [2]. these SYN packets typically outnumber other TCP packets many times over. There is a paper available on the Cisco web site titled “Characterizing and Tracing Packet Floods Using Cisco Routers”. 6. Still. the upstream provider to the victim may be able to filter out all ICMP echo replies to the target network.

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

txt) RAT Example The transcript below shows a small RAT session (version 2. ncat_report: writing all.. (e.g.1c 259 .txt.conf/ ncat_report: writing northconf-2aug03.conf ncat_config: c:\CIS\RAT/etc/configs/cisco-ios/cis-level-2. ncat_report: writing rules.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 . Version 1.ncat_report.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-2.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-1.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-1. 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. D:\routeradmin> set PATH=c:\CIS\RAT\bin.Testing and Security Validation • A text file that contains IOS command to fix the problems identified in the report. ncat_report: writing northconf-2aug03.txt.1) run on Windows.html. Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/common.html. ncat_report: writing all.%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.txt. .conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/local. ncat_report: writing northconf-2aug03.ncat_fix.ncat_fix. ncat_report: writing all.html).html (cisco-ios-benchmark. D:\routeradmin> Note that this example does not show all of the questions posed by ncat_config..conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/local.ncat_fix. .txt. Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/common.conf ncat_config: c:\CIS\RAT/etc/configs/cisco-ios/cis-level-1.conf/ Parsing: /c:\CIS\RAT/etc/configs/cisco-ios/cis-level-2. routerconf-1.conf/ Checking: northconf-2aug03 done checking northconf-2aug03.ncat_report.

treat it as a potential issue. do not follow them blindly. Examine each rule failure that RAT reports. your router must meet your local security policy. In the end.Router Security Configuration Guide RAT Caveats Use the RAT benchmark reports as an aid in securing your routers. RAT is simply a useful tool to help you find problems and areas for improvement. 260 Version 1.1c . and check the router settings related to the rule carefully.

html A good summary of basic IOS router security practices. CERT Coordination Center. [2] “Characterizing and Tracing Packet Floods Using Cisco Routers”.html Version 1. Cisco Technical CERT Coordination Center.1c 261 . 1999. CERT Incident Note IN-99-07. [6] “Distributed Denial of Service Tools”. Cisco Systems. Cisco White Papers. available at: http://www. available at: http://www.5. available at: http://www. available at: http://www. “The Latest in Denial of Service Attacks: Smurfing”. a web-based tool that registered Cisco customers can use for a cursory security analysis of their IOS configurations. 2000.html [5] “Denial of Service Attacks”. available at: http://www. Cisco White Papers.html [4] “Defining Strategies to Protect Against TCP SYN Denial of Service Attacks”.cisco. available at: 1996. Software Engineering Institute.html [8] “Distributed Attack Tools”. available at: http://www. also includes a reference to the Cisco IOS Output [9] Huegens. Software Engineering Institute. 1997.html [7] “Topic: TCP SYN Flooding and IP Spoofing Attacks”.org/incident_notes/IN-99-07. Securify Cisco Systems.governmentsecurity. 2000. Packet Storm. 1999. CERT Coordination Center. CERT Advisory CA96. available at: 2000.cert. Software Engineering 2000. available at: http://www.cert. Cisco Technical Cisco Systems.html A good overview of DoS attack C. Cisco Systems.Testing and Security Validation 6.html Detailed guidance on tracing certain kinds of DoS attacks. References Web Sites and On-Line Resources [1] “Improving Security on Cisco Routers”. [3] “Strategies to Protect Against Distributed Denial of Service (DDoS) Attacks”. NESSUS .securityfocus.1c .net/find_products/ http://packetstormsecurity. 2003. Security Configuration Guide [10] SAINT http://www. available from: 262 Version Additional Exploit-Related Pages: Additional General Exploit Information Sites: http://www. M.html ISS Internet Scanner Automated security scanning and testing tool sites: RAT http://www. “Exploiting Cisco Routers (Part One)”.html SATAN http://www.

while others require a real router. • Protecting physical access to the switch itself • Controlling virtual access to the switch.1.Additional Issues in Router Security 7. 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. 7. 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 diagram below shows how virtual LANs can be set up to emulate two physical LANs spread across two switches. while removing the ability to impose some security restrictions. Therefore.3). Routing and Switching As network bandwidth demands continue to increase.1c 263 . Switches typically have limited or negligible filtering capabilities. Note that some switches can act as routers between their separate VLANs. routers can supply critical protection at network boundaries by filtering traffic (see Section 4. Version 1. For example. each individual host and device must be configured securely rather than relying on protection at their LAN boundaries. 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. IP routing will increasingly be replaced by layer 2 switching in high-performance applications. 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. in a network environment that is predominantly based on switching. One feature of switched environments that might be usable to improve security is Virtual LAN switching. The list is not comprehensive.

2) Host X (14.4) Host Z ( 264 Version 1.2.2. but it is clear that VLAN security will grow in importance in the next few years.2.2.3) Host C ( Host Y (14.3.3) LAN 1 14.6) Virtual View Host B (14.4) Host Z ( Host X (14.7) Host B (14.0/24 Host Y (14.3.8) Internet Real Construction Router LAN Switch 1 LAN Switch 2 Host A (14.7) Host C (14.Router Security Configuration Guide Internet Host A ( Router LAN 2 . Figure 7-1: Virtual LAN Switching More investigation is needed to determine the security roles and policies for configuring VLANs.2.

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. For more information. 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). In the mid1990s. 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. The Internet community is gradually migrating from IPv4 to v6. and later. Version 1. The supplement will be available from www. IP version 6 (IPv6) was first standardized in 1995. By mid-2005. Defining best practices and configuration guidelines for routers handling IPv4 and IPv6 may be complicated. and the IETF initiated efforts to standardize a successor. The US government and US DOD have set ambitious goals to complete substantial migration by 2008.2. 12. IPv6 The Internet was built using version 4 of the Internet protocol (IPv4). an IPv6 supplement to this guide is in preparation. such as [16]. 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 translation. and other protocols 576 bytes optional. The table below gives a few comparisons between IPv4 and IPv6. consult any good book on Security best practices for IPv6 are not yet well-established. not widely used 128 bits usually fixed. and should be published by mid-2006. due use of transition mechanisms and interactions between IPv4 and IPv6 features. ICMP. At the time of this writing.2T. although some recommendations exist (see [15]).nsa. and the standard revised to its current form in 1998. it became clear that the Internet would eventually outgrow the address space and capabilities of IPv4. hundreds of IPv6 address blocks had already been allocated to network carriers and enterprises all over the world. During the transition period.Additional Issues in Router Security 7.1c 265 .3.

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

In an MPLS network.Additional Issues in Router Security 7.4. and Virtual Private Networks (VPN) capabilities. 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. incoming packets are assigned a label by a label edge router (LER). the current version is described in RFC 2547 and RFC 3031. Version 1. offering traffic engineering (TE) . Each label switch router (LSR) makes forwarding decisions based only on the contents of the label. the LSR removes the existing label and applies a new label which tells the next hop how to forward the packet. Quality of Service (QOS). MPLS is defined and specified by the IETF. MPLS enhances the services that can be provided by IP networks. At each hop. Packets are forwarded along a label switch path (LSP). MPLS uses label switch technology to simplify routing and enhance overall network performance. Label switching allows routers to make forwarding decisions based on the contents of a simple label.1c 267 . rather than by performing a route table lookup based on destination IP address.

Installing certificates and setting up authorities on Cisco routers is complex but well-documented. Currently. IPSec and Dynamic Virtual Private Networks Section 5. depending on the installed IOS version and feature set.509 Certificate Revocation Lists (CRLs) and the Online Certificate Status Protocol (OCSP). To achieve scalability.0 Network Security [2]. As deployment of VPNs becomes more common. However. There are also trust issues in any large VPN deployment: are all members of the VPN trusted equally? In general. including X.5. but not for authorization or access control.2 explains some of the basic features of IPSec. consult Cisco IOS 12.2 probably will not scale well enough to satisfy users’ needs. • Certificate revocation – In any large-scale PKI. support the ability to protect certain traffic based on its protocol and port numbers. 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. For complete information on the IPSec and dynamic VPN capabilities of Cisco IOS 12.1c . and it must have its own certified public key and private key. • Cryptographic issues – Selection of uniform key sizes and cryptographic algorithms will be a contentious issue in VPN deployment. • PKI enrollment and obtaining certificates – To participate in a dynamic VPN based on Public Key Infrastructure (PKI). Cisco routers currently support only a small complement of algorithms.Router Security Configuration Guide 7. It may be necessary to configure revocation checking on routers participating in dynamic VPNs. IPSec and Virtual Private Network (VPN) configuration are complex topics. the simple configurations described in Section 5. removing certified principals from the trusted community is very important. 268 Version 1. VPNs will need to be dynamic. PKI standards define various data formats and protocols for defining revocations and for checking certification status. and most other VPN systems. IPSec is most useful for integrity and confidentiality assurance. there are no uniform guidelines for selecting traffic to protect. • Designating traffic to be encrypted – Cisco routers. a router or any other device must possess a copy of the correct root and authority certificates.0.

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. expect a corresponding increase in mobile users and addressable devices expecting to join home base networks.4.2. such as Ethernet or PPP.6.0/24 Remote User 14. the primary such protocol used today is called the Layer 2 Tunneling Protocol (L2TP). There are several approaches to doing this.9.185 Figure 7-2: Overview of Virtual Private Dial-up Networking In general. and protected networks from remote sites.2. 7.29 Internet home base router dial-up File Server 14. like PPP. This kind of tunneling architecture is called Virtual Private Dial-up Networking (VPDN). over IP from a remote router or network access server to a central router. but it seems possible that small deployments could use static IPSec tunnels as described in Section 5.19.9. and actually use their home base LAN address. VPNs. Virtual Private Dialup Networking Cisco routers support tunneling dial-up protocols. each of which has different security issues.2.1c 269 .1. Standard protocols exist for tunneling layer 2 protocols.10 modem Virtual Connection Inside LAN 14. over IP networks. from a remote part of the network. and it is illustrated in the figure below. Version 1.6. This is an area that needs further study.Additional Issues in Router Security 7.9. Tunneling Protocols and Virtual Network Applications As VPNs become more popular and widespread. Use of such tunneling protocols allows remote users to join a LAN. Virtual Private Dial-up Networking configured between these two devices Access Server 126.2.

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

1c 271 . the domain name resolver in Cisco IOS cannot recognize or check DNS security extensions). Given the negligible deployment that secure DNS has enjoyed in the first couple of years that it has been widely available. These security services are supported by the latest versions of the primary Internet DNS server implementation. Cisco routers do not yet support acting as a secure DNS client (in other words.250).Additional Issues in Router Security to IP addresses (like 14. it seems unlikely that it will become ubiquitous. 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. 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 basic DNS protocol offers no authentication or integrity assurance. The DNS Security Extensions standard defines comprehensive integrity and authentication facilities for DNS.mydomain. Bind. the mapping between names and addresses is fully authenticated and integrity assured. It could then perform the security checks on remote DNS requests. Version 1. In a network with secure DNS. Secure DNS The Domain Name System (DNS) used on the Internet and other major networks provides the mapping between names (like central. and pass along only validated results.2. In a network with secure DNS.9.

The updated standard for secure DNS. This documentation volume includes extensive configuration information for Cisco ATM switching and LANE. [5] Kent. including IPSec and related configuration commands. “RSVP Cryptographic Authentication”. defines the protocol structure and intent. includes extensive remarks about VPN architecture. IN. and Jamin.0 Switching Services. and Atkinson. R. 1998. includes extensive discussion and examples. Contains a good overview and substantial technical detail about IPSec and related topics. 1999. and Virtual Private Networks.. 1999. “Domain Name System Security Extensions. and Harkins. 1999.0 Network Security.9. [4] Doraswamy.” RFC 2401. G. M. 1998. NY. Indianapolis. 2000. J. Upper Saddle River. N.. Cisco Press. “Resource reSerVation Protocol (RSVP) – Version 1 Functional Specification. and Talwar. Cisco Press. Describes the message authentication service to be used with RSVP. Prentice-Hall. 1997. NJ. The basic standard for RSVP. 2000. Authoritative source for in-depth descriptions of security-related IOS facilities. [3] Cisco IOS 12. [2] Cisco IOS 12. References [1] Sacket.” RFC 2205.C. RFC 2225. McGraw-Hill. Z.” RFC 2535. H. RFC 2747. S. 1999. The definition of Classical IP over ATM. Contains a good overview of Cisco ATM facilities. [7] Braden. The master document for IPSec. Berson. D. and Halpern. Indianapolis. also good background reading for understanding the issues of hosting IP over ATM.1c .Router Security Configuration Guide 7. “Classical IP and ARP over ATM”. [8] Baker. [6] Eastlake. “Security Architecture for the Internet Protocol. New York. IPSec: The New Security Standard for the Internet. [9] Laubach. Cisco Router Handbook. D. Intranets. Lindell. 272 Version 1. IN.

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

daytime) based services.2] • Small services . discard. Keep the offline copy of each router configuration in sync with the actual configuration running on the router.1] Implement access list filters by permitting only those protocols and services that the network users really need.3] Run the latest available General Deployment (GD) IOS version.1c . along with some important general principles for maintaining good router security. [Section 4.miscellaneous UDP (echo.5. [Sections 4. [Section 4. Use the no cdp run command to kill the process and disable CDP globally. General Recommendations Comment and organize offline editions of each router configuration file! This sounds fluffy despite being a big security win. Some services that should almost always be turned off are listed below. 8. chargen. To leave CDP running but disable it for certain network connections. [Section 4.Router Security Configuration Guide 8.5. and keep all it and all old versions under configuration management. apply the command no cdp enable to the appropriate interfaces. for more details consult the corresponding parts of Section 4. One of these is the UDP echo which is used in the ‘fraggle’ attack. discard. This is invaluable for diagnosing suspected attacks or problems and recovering from them. chargen) and TCP (echo. Shut down unneeded services . [Section 4.1. 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 commands no service udp-small-servers and no service tcp-small-servers to turn these off. then turn off clearly unneeded facilities and services. and save memory and processor slots too. • CDP . References to appropriate parts of Section 4 appear at the end of each recommendation. Start by running the show proc command on the router.things that aren’t running can’t break. Appendices The sections below offer ancillary material and supplemental guidance for network and security administrators. 8. and explicitly denying everything else.3] Specific Recommendations 1. Trying to deny just the ‘bad things’ is a losing proposition. CDP sends packets from your router once a minute or so identifying your router. The descriptions here are terse.2] 274 Version 1.Cisco Discovery Protocol is used almost exclusively by Cisco RMON.

9. If the router has an auxiliary console port (aux port) and it is not in use. Turn off this little-used server with the command no ip bootp server. [Section 4. If packets are rejected by an access list filter for other reasons.the finger daemon. the log message will often list the packet as using “port 0”. 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. turn it off with no ntp server.2] • NTP . Don't be a Smurf buddy! While the Smurf attack doesn't usually attack the router itself. Log access list port messages properly.2] Central(config)# interface eth 0/0 Central(config-if)# no ip directed-broadcast 3. but include it in your configuration explicitly anyway. For reasons of efficiency. [Section 4.1] 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 4.6 East(config)# access-list 51 deny any log 5. This may be the default on some recent versions of IOS.2 and earlier) or no ip finger (IOS 11. [Section 4.Appendices • Finger . use the special port range arguments shown below. use the command no ip directedbroadcast on all interfaces.3 and later).the Network Time Protocol. To prevent this from happening. 4. Use the command no service finger (IOS 11. [Sections 4. the attacks will appear to come from you. [Section 4. Always start an access-list definition with the command no accesslist nnn to make sure it starts out clean. Check them with the show ip interface brief command. If NTP is not being employed for time synchronization.2. shut it down as shown below.2] 2. a Smurf attack can let an attacker use your network to launch denial of service raids on other sites. no access-list 106 access-list 106 deny udp any range 0 65535 any range 0 65535 log Version 1. To prevent this.1c 275 . [Section 4. Cisco IOS doesn't look at an entire packet header unless it has to. NTP can also be disabled for only a specific interface with the ntp disable command.2.3] East(config)# no access-list 51 East(config)# access-list 51 permit host 14.5] • BOOTP – the IP bootp server.

Allow only internal addresses to enter the router from the internal interfaces. such as Telnet or SSH.0.0 0. Turn off SNMP trap authentication to prevent a remote SNMP system shutdown request. [Section 4. In IOS 11.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. use the transport input command to enable only those methods.1c .5. use an access list and the access-class command.0.2. enforce this using access lists. If only specific methods of attaching to the VTY. If SNMP is not being used on the router. turn it off with the command no snmp-server. it helps identify mis-configured internal hosts and 276 Version 1. By default. [Sections 4.2 and later use the global configuration command no snmp-server enable traps.3] South(config)# no snmp-server enable traps South(config)# no snmp-server 9. [Section 4.3] 6. [Section 4. Password and access protect the Telnet VTYs. Unless the network is one of those very rare setups that needs to allow source routed packets. [Section 4.1] 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 14.10. Block illegal addresses at the outgoing interfaces.1] South(config)# enable secret 2Many-Routes-4-U South(config)# 10.1 and 4.6 for guidance.2] Central(config)# no ip source-route 8. 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. 7. Besides preventing an attacker from using the router to attack other sites. are permitted. use the password command.2. consult Sections 4. To set a password. the source routing facility should be disabled with the command no ip source-route. virtual terminals (telnet) are unprotected.255 Controlling authentication for login to the router is an extremely important topic. 4.

• If the network does not need IP multicast then block it.0. Block packets coming from the outside (untrusted network) that are obviously fake or are commonly used for attacks. Incorporate this protection into the access lists applied to interfaces connected to any untrusted networks. These packets cannot be real. untrusted network. (A superior but more difficult approach is to permit only necessary ICMP packet types.1).9.0.255 any log Version [Section 4. • Block broadcast packets.1.6 12.255 any log ! block bogus loopback addresses access-list 107 deny ip 127. Turn on the router’s logging capability. 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 14.0 0. This protection should be part of the overall design for traffic filtering at the router interface attached to the external. [Section 4. Make sure that the router blocks syslog traffic from untrusted networks.Appendices networks. [Section 4.255 any log access-list 107 deny ip 14.0.1c 277 . and use it to log errors and blocked packets to an internal (trusted) syslog host. (Note that this may block DHCP and BOOTP services.0.0 0. but these services should not be used on external interfaces.) The example below shows how to enforce these rules on router North.0.0.0.) • A number of remote attacks use ICMP redirects.3] East(config)# no access-list 101 East(config)# access-list 101 permit ip 14. This impedes some TCP sequence number guessing attacks and related attacks.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 11.3] • Block packets that claim to have a source address of any internal (trusted) networks.0.0.0] Central(config)# Central(config)# Central(config)# Central(config)# logging logging logging logging buffered trap info facility local1 14. block them.2. This approach may not be feasible for very complicated networks. • Block incoming loopback packets (address 127.

255 any North(config)# ! block broadcast North(config)# access-list 107 deny ip host 0. a ‘Land’ attack on the router itself).255.0 0. disable network autoloading with the command no service config. using the command service tcp-keepalives-in.Router Security Configuration Guide North(config)# ! block multicast North(config)# access-list 107 deny ip 224.5] East(config)# service timestamps log date \ msec local show-timezone East(config)# 18.2] Central(config)# interface eth 0/0 Central(config-if)# no proxy-arp 16. use the global configuration command service timestamps as shown in the example below. Disable it by applying the command no proxy-arp to each external interface.1c . [Section 4.0. however. To be safe. using a rule like the one shown below (part of the configuration file for router East). If proxy ARP is not needed.0.2.2] 278 Version 1.6. Block incoming packets that claim to have the same destination and source address (i.0.250 255.3] no access-list 102 access-list 102 deny ip host 14. [Section 4. Unless the router absolutely needs to autoload its startup configuration from a TFTP host. include the command no ip http server in all router configurations. disable it on all interfaces. .0.2. So that the complete date and time are stamped onto entries in the routers buffered log.0 ip access-group 102 in 14. the HTTP server is off by default.250 log access-list 102 permit ip any any interface Eth 0/0 ip address 14.0 any log North(config)# ! block ICMP redirects North(config)# access-list 107 deny icmp any any redirect log .255.255.e.2. Incorporate this protection into the access list used to restrict incoming traffic into each interface.6. [Section 4. [Section 4. North(config)# interface eth 0/0 North(config-if)# ip access-group 107 in 13. [Section 4.250 host 14.2] 17.1] 15.0. Proxy ARP is used to set up routes on the fly for internal hosts or subnets and may reveal internal addresses. Turn on TCP keepalive packets for administrative telnet sessions.6. Except on the rarely-seen Cisco 1000 series routers. [Section 4.

Turn on password encryption. It is not necessary to install each and every new IOS release. Installing an update normally imposes some downtime. [Section 4. but it is a good idea to keep your router up to date.Appendices 19. see Section 6. This provides some security against casual ‘over-the-shoulder’ attacks.1c 279 . and defending against common attacks.5] For more information about testing router security. In general. newer releases will include fixes for security bugs. so plan your updates carefully. so that regular passwords are stored and displayed in scrambled form. and will provide new security features.1] East(config)# service password-encryption 20. Update your IOS image to the latest General Deployment (GD) release. [Section 4. Version 1.

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

2. If you do not have the switch documentation handy. Catalyst 5000 series). and thus does not need to be configured. While the command syntax and command interface structure differ between Cisco IOS-based and other equipment. and configuration loading and maintenance. RELEASE SOFTWARE (fc1) . the operating system name and version are underlined in the examples below. Most of the recommendations in this section apply to switches. 51 minutes Cat5k# The table below describes how to apply the guidance in each part of Section 4 to IOS-based LAN switches. Uptime is 45 days. Version McpSW: 4.g. 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.2(8)SA. In addition.g.2. 2900 series) and those that do not use IOS (e. but they can be divided into two broad groups: those that use Cisco IOS or a derivative (e. . Especially important for 2900 switches is restricting access to the HTTP server.2 Network service security Version 1.1(2) . 4.Appendices 8.5(1) . check whether the particular switch is running IOS or some other operating system. login to the switch and use the show version command to display the operating system name. . all ports should be configured to block traffic to unknown addresses using the port block interface configuration command. . Application to Cisco IOS-based LAN Equipment Cisco makes several kinds of network switches. 3 hours. 3 hours. Much of the security guidance given in Section 4 that can be applied to IOS-based Cisco switches. Version 11. Before attempting to apply the detailed instructions from Section 4. System Bootstrap Version 5. . IOS-based Catalyst 2900 sw20c# show version Cisco Internetwork Operating System Software IOS (tm) C2900XL Software (C2900XL-H-M). The syntax shown in Section 4 will work for IOS-based switches.1 4. and even some smart Ethernet hubs. but will not generally work on other devices. sw20c# Non-IOS Catalyst 5500 Cat5k# show version WS-C5505 Software.1c 281 . 9 minutes . any network service that is related to routing usually is not supported on a switch. the same general principles apply to all of them. remote access restrictions.5(1) NmpSW: 4. sw20c uptime is 6 days.

It should be configured using Section 4 like any other router. too. All switches support RMON and SNMP. VLAN configuration. but not for filtering traffic passing through the switch. although some Cisco switches can act as routers. if they support AAA (IOS This highly detailed guide covers general switch security. 4.. available under: http://www. and must have their time set manually.4 Routing protocols 4. Duesterhaus. but do not use them for as many different purposes as a router does. December 2000. GIAC Paper. includes several good references. and more. available at: http://www. et al. on a switch. after careful consideration of its role in the network security policy. References [1] Turner. This module is essentially a 4700-series IOS router attached to the switch.synfin.nsa. “Cisco IOS Switch Configuration Guide”.3. some switch IOS versions do not support NTP.g.1c . NSA.Router Security Configuration Guide Section Topic Access lists Application to Switches IOS-based switches support IP access lists.3 4. or access to them should be restricted.5 Audit and Management 4.2 and later). Consult [2] for more detailed guidance and specific examples.] Almost all of this section applies to IOS-based switches. A.pdf An examination of the security (and lack of it) provided by separating traffic with Ethernet switches. June 2004. Catalyst 5000 series Route Switch Modules) can and should be configured using the guidance in Section 4. Basically.2.6 Access control with AAA Note that Cisco switch-resident routing hardware ( but the discussion is still valuable. [2] Borza. “Network Insecurity with Switches”. 282 Version 1. these services should be disabled if not in use. All of this section is applicable to IOS-based switches. This section is not usually applicable to switches. [Note: some Catalyst 5000 and higher series switches are equipped with a ‘Route Switch Module’. 8. Many of the issues discussed in the paper have since been fixed. access lists are used for limiting access to services on the switch itself. Most of the security testing guidance given in Section 6 also applies to LAN switches. port management.

revision 3. There is a complex naming scheme for early releases that is beyond the scope of this guide. and may also include a number in parentheses. For operational purposes. such as 12. “F”. consult [1] for complete details.1.3T” is IOS version 12. 11. especially during the flurry of bug fix releases that tends to follow a major change. 11. For example. You might also see maintenance revision numbers in parentheses.1. 12 Examples: 12. “HA”. 8. Some of the suffixes that you might see on special-purpose releases include “XA”. for example. operational routers should kept up to date with the newest stable release that provides all the needed features. to understand what Cisco provides. unless it has some required. release number and release identifiers tell what features could be available. Unfortunately. identifier T. It is intended to help with upgrade strategies and version selection.0. Often it will not be practical to install all the updates that Cisco makes available. and the release naming scheme. The current format for a Cisco IOS release name is shown below. Release Levels and Names Cisco follows strict naming schemes for IOS releases. and then follow it. and the revision number tells how many times the release has undergone fixes to correct problems. the format has changed several times since IOS was first introduced in the mid-1990s. the release “12. Overview of Cisco IOS Versions and Releases Cisco provides a very large number of software releases for their routers and other products.Appendices 8. This appendix provides an overview of the major release levels. Version 1.0 Revision = 3 Release = 11.1.3 Revision = 5 Identifier = T 11.3. critical feature.3.0. usually for ED releases. Devise a consistent upgrade strategy that matches the needs of your network. Cisco releases may be broadly divided into kinds: regular shipping releases (general or limited) and early releases. it is usually best to avoid early release software.N.3.1c 283 .M RR Release identifier Maintenance revision number Minor release number IOS Major release number: 10.3 Release = 12.2(9)XA. use this appendix and the materials listed in the references.5T Figure 8-1 – Cisco IOS Release Naming In general. The ‘T’ identifier designates an early release of new technology features. VV. A regular release will almost always have a simple number with no release identifier.8. In general. An early release will usually include an identifier.

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

12.1 release was the last IOS release to use the old ‘classic’ or monolithic architecture. • Named access control lists (See Section 4. Cisco also offers a hardware/software compatibility matrix checker. and 12. The lists below describe some of the major features introduced into IOS in each of these releases. IOS 11.1. and engineering development for it was dropped in 1999.3) • Network address translation (NAT) • Support for RSVP and IP Quality-of-Service (see Section 7.5) • The IOS web server and web browser management interface [IOS 11. All earlier Cisco IOS releases.1.3. Major Releases and their Features There are at least five major releases of Cisco IOS software currently in use in operational environments: 11.5) • Various OSPF and BGP4 enhancements Version 1.1c 285 .1 was first deployed in 1996. are now mostly unsupported by Cisco. Some of the important features • RIPv2 (see Section 4. 3660). 11. 11.2. 7500. 1605. with emphasis on security-relevant features.0. and earlier.1 The release was the first IOS version to fully implement Cisco’s modular architecture for router software. but is not available for some of the popular newer models (e.2 The 11.1(5) and later] • RADIUS support (as part of AAA. freely available on their web site. 12. a few of them are listed below.Glossary which you hope to run the software. 8.3. A great many new features were added to IOS over the lifetime of 11.2. IOS 11. see Section 4. IOS 11. you can check what IOS releases are supported on your router model. it did not offer extensive security features.6) • RMON support (see Section 4. Using this tool [3].2.1 is available as a GD release for all older Cisco routers. 12. While exceedingly stable and robust. although some of them are still available for download.g.5) • Lock-and-Key dynamic access lists IOS 11. Ensure that amounts of installed memory meet or exceed the requirements before attempting to load the IOS release.

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 11.2 is available as a GD release for many older Cisco router models.4) IOS 11.0 and 12.0T releases brought together a wide variety of features that had previously been available only in selected LD and ED releases of IOS 11. • 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. but it was never officially shipped as a GD release. 12. Some of the features introduced in 11. but only at the ED and LD release levels.3 was used to introduce a large number of new features into IOS.3T) • Cisco Encryption Technology (CET) VPNs • Enhancements to AAA (See Section 4.0 The 12.0 are listed below. and to help eliminate the confusion of specialized releases that plagued 11.3.3 are listed below. IOS 12.Router Security Configuration Guide • Initial support for TCP Intercept (IOS 11.0 was designed to be the basis for future router software releases. • Initial implementations of IPSec (11.1c . Some of the security-relevant features introduced or consolidated in 12.1 through 11.7) • Full IOS firewall feature set and CBAC (11. IOS 11.2F only) • Early (pre-IPSec) VPN support • Early versions of the IOS Firewall feature set and CBAC (see Section 5.3 11.3 is available for almost all Cisco router models.

A few of the many enhancements in 12. • 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 12.6) • Time-based access lists • General availability of IP unicast reverse-path verification [Section 4.0. • Enhanced IPSec certificate management and AAA integration • AAA enhancements: server groups. and also expanded the availability of some features.3 are listed below. IOS 12. as well as enhancements to some core security features.2 The 12. IOS 12. and many other Cisco hardware products.4] IOS 12.3 release added several substantial new security features. it was the latest release available at GD status for some devices.2 reached GD status in 2002.3 had not yet reached GD status. As of summer 2005. simplifies basic secure configuration • New features added for password and authentication controls Version 1. As of the summer of 2003. • Secure Shell Version 1.1 The 12.Glossary • SNMP version 3 (See Section 4.0 is available in both LD and GD forms for almost all supported Cisco router platforms.1.1 release is an incremental step forward from 12. Some of the security features that appeared in 12.1c 287 .2 release adds some new features to 12.5 (SSHv1) supported in nearly all IOS editions • Cisco Auto Secure feature added.1 are listed below.2 are listed below. IOS 12.3 The 12. Some of the security updates in 12.

cisco.. available at: http://www. The service is available to registered customers at http://tools. (Note: may be available to registered users only. R.0. Murphy.3.. [2] “Cisco IOS Reference Guide”. 2000. client and server. • 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 • Secure Shell Version 2 (SSHv2) supported. 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.1c . 2001. Cisco Systems. CCIE Professional Development Series. C.3. References [1] Inside IOS Software Architecture. M.html This detailed web page explains the IOS release naming scheme. Cisco IOS Releases: The Complete Reference.x/Support/HWSWmatrix/hwswmatrix. A few of the new security features are listed below.cgi This interactive web page allows you to find IOS releases compatible with particular router models.M. Cisco Press. Cisco Systems..2. 2000. [3] “Hardware/Software Compatibility Matrix”. Cisco Press. Cisco White Papers. V.3. it only covers up through IOS 12. focused on packet handling and routing. 2004. 288 Version 1. 8. and includes a map of releases up through 12. This highly specialized book covers the Cisco IOS release system and release history in painstaking detail. available at: part of the Cisco Software Advisor. decribes features in IOS 12.4 release adds additional security features and many new protocols and functions to IOS.) [4] Bollapragada. White.4 The 12.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. A very detailed examination of However.jsp.

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

ICMP is standardized in RFC 792. Exterior Gateway Protocol – routing protocol designed for managing route updates between different autonomous systems. Domain Name System – hierarchical naming scheme used for host and network names on most IP networks. Extended Interior Gateway Routing Protocol – A Cisco proprietary routing protocol that includes peer authentication features. or that crash it. like IP addresses and gateways. derived from the command used to request privileged EXEC mode: enable. Denial of Service – this abbreviation is often used for network attacks that prevent a network component from providing its operational functions. FTP control messages are passed on TCP port 21. ICMP is a network layer protocol that provides error messages and management capabilities in IP networks. IKE.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. A slang expression for a privileged EXEC session on a Cisco IOS router. FTP is standardized in RFC 959. AH) File Transfer Protocol – widely-used TCP-based file transfer and file management protocol. DNS is also the name for the protocol used to transmit and relay name information. The main EGP in use today is BGP version 4. Encapsulated Security Payload – a part of IPSec. DNS DoS DDoS EGP EIGRP Enable mode ESP FTP GTSM ICMP 290 Version 1.1c . the packet format and protocol for IP confidentiality services (see also IPSec. Generalized TTL Security Mechanism – a simple spoof rejection mechanism that uses the TTL field of the IP header to detect illicit packets. Internet Control Message Protocol – a support protocol used along with IP for control and status message. DNS is standardized in RFCs 1034 and 1035. Dynamic Host Configuration Protocol – UDP-based protocol for assigning host network attributes. including the Internet. Typically. GTSM is standardized in RFC 3682. (see also OSPF). DHCP is standardized in RFC 2131. on the fly. 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.

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

Multi-Protocol Over ATM – A proposed standard mechanism for hosting network protocols (such as IP) over ATM. NTP is standardized in RFC 1305.0. Media Access Control address – the link layer address of a network interface. standardized in RFC 1321.0 to 225. can use UDP or TCP.255 are usually multicast group addresses. certificates. Network Time Protocol – the standard network time synchronization protocol.255. port 123.255. Cisco routers use NVRAM to hold their startup configuration. Management Information Base – the hierarchical data organization used by SNMP. especially Ethernet interfaces. An application that acts as an intermediary in the network exchanges between applications or services. Proxy servers are often employed to moderate exchanges through a firewall. Network News Transfer Protocol – a TCP-based application protocol that usually runs on port 119. addresses from 224. L2TP is standardized in RFC 2661. (See also SNMP) Multi-Protocol Label Switching – a standard mechanism for transferring packets over backbone networks by tagging them with labels. standardized in RFC 3031.Router Security Configuration Guide L2TP Layer 2 Tunnel Protocol – A standard protocol for forwarding lowlevel protocols over IP networks. but usually uses UDP. (See also RIP. in which packets can be broadcast to particular recipients based on address. IGP.1c .0. and enrollment. OSPF is standardized in RFC 2328. Open Shortest Path First – an IP routing protocol that uses a linkstate distance metric. In IPv4. An operational feature of IP. MAC Address MD5 MIB MPLS MPOA Multicast NNTP NTP NVRAM OSPF PKI Proxy 292 Version 1. An Ethernet MAC address is 48 bits long. Non-volatile RAM – device memory that can hold data even when unpowered. Message Digest algorithm 5 – a widely-used cryptographic checksum algorithm. EIGRP) Public Key Infrastructure – mechanisms and components for management of keys.

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

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


fi/~walker/minicom.kiwisyslog. Nessus The FreeRADIUS server is a highly configurable open-source RADIUS server implementation. minicom is fast. Tool References The list below describes some available commercial and non-commercial tools that may be helpful in router administration and improving network security. flexible.pp.freeradius. efficient. Tools like Ethereal are valuable for diagnosing and testing router and network security. a network traffic capture and analysis The Nessus security scanner is a handy tool for getting a quick idea of the security vulnerabilities present on a A syslog server is necessary to capture and preserve log messages from Cisco routers and many other network devices.3. While it has no fancy NET-SNMP is a free software toolkit for SNMP. earlier versions should not be upgraded.nessus. originally created and distributed by the University of California at Davis. Kiwi Syslog http://www.1. While Nessus is primarily oriented toward scanning host computers. It was formerly called “ucd-snmp”.1c . and can serve as an effective Cisco router console application on Linux.html Minicom is a small. NET-SNMP http://net-snmp. effective terminal emulation tool for Linux and Unix. The Kiwi Syslog is one of several freely available syslog servers for Windows operating systems. it may also be used to scan Ethereal is an effective “sniffer”.Router Security Configuration Guide 9.sourceforge. 300 Version 1. FreeRADIUS http://www. Minicom http://www. Ethereal http://www.clinet. The current version is 0.ethereal.9.

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

257 smurf Attack. 166 show rmon. 98 Conventions. 75 copying. 228 crypto maps. 150. 168 show line. 209 Logging and. 119 Accounting. 276 and VTYs. 127 Banners. 59 Application Layer. 97. 110. 28. 157 show running. 66. 25. 128 cdp. 83 banner. 160 show ip. 211 Audit. 196 resolution. 79. for IPSec. 228 debug. 73 ip inspect. 93. 91. 75 service tcp-keepalives-in. 79 ip ospf. 77. 68. 239 Abbreviations. 60 mop enabled. See Logging Authentication for IKE/IPSec. 198 key chain. 177. 184 aaa authentication. See Routing. 29 DDoS. 77 ntp. 88 policy-map. 173 distribute-list. 127 TTL Security. 184 authentication. 252 Console. 180 authorization. See AAA timeout for SSH. 112 router isis. 126 route flap damping. See Users Address CIDR allocation. 183 Kerberos. 187. 233 ARP. 176. 275 SYN flood. See also AAA Accounts. 60 username. 194 BGP. 75 service pad. 193 access-class. 253 testing with. 125 router eigrp. 181 aaa authorization. 67 configuring. 115 kerberos. 114 router ospf. 72 class-map. 237 ip name-server. 96 CBAC and. 57. 168 Committed Access Rate (CAR). 142. 98 radius-server. 161 snmp-server. 145 mirror. 215 DMZ. 195 TACACS+. 161 ip http. 72 service-policy. 183 aaa new-model. 123 prefix filtering. 266 Attack Tests. 208 retries. 278 table. 229 servers. 117. 207 exec-timeout. 98 clock. 59 logging. 252 Land attack. 95 rmon. 75. 228 ip verify. 72 CEF. 60 access-list. 180 aaa server group. 59 bgp dampening. 132 CDP. 60 hostname. 18 Debugging. 110 service config. 163 shutdown. 72 service udp-small-servers. 35. 10 Access Control Lists. 239 and access lists. 78 Secure DNS. 145. 253 Denial of Service. 206–11. 83 URPF and. 126 spoofing. 10 CPU. 154 show version. 170 ATM. 131 CIDR. 56. 60 logging. 99 show crypto. 95 extended. 95 Configurations auto-loading. 67 show snmp. 157 router bgp. 179 RADIUS. 265 proxy ARP. 178. 215 named. 239 viewing current. 84 IPSec and. 15 martian. 109 router rip. 172. 192 server groups. 58 302 Version 1. 21 Commands aaa accounting. 132 isis authentication. 147 config t. 62 encryption. 90 password sniffing. 150 passive-interface. 237 testing. 180. 83–93. 242 ip cef. 271 SSH and. 194 terminal. 37 Domain name service Kerberos and. 61 show logging. 60 reacting to. 60 CAR rules.1c . 72 control-plane. 228 D Dampening. 252 Attacks. 79 ip audit. 66 service tcp-small-servers. 25. 227 Authorization. 142–45 login authentication. 68 examples. 79. 143 trace ip. 181 login local. 252. 124 Black Hole Routing. 98 Cryptography. 22 AS numbers. 195. 22 IP. 236 global parameters. 111 ip ssh. 193. 127 ip rip.Router Security Configuration Guide Index A AAA. 132 ip ftp. 154 tacacs-server. 236 compiled. 88 enable secret. 112 line. 206. See AAA Auxiliary port. 99 copy. 125 transport. 59. 15 MAC address. 193 rate-limit. 57 connect. 105. Null routing C CBAC. 175–200 accounting. 41. 160. 109 ip prefix-list. 195–200 method lists. See BGP route flap damping Data Link Layer. 212 show flash. 140 Control plane policing. 84 standard. 60 B Backbone Router. 161 crypto. 50.

217 rules for. 113 OSPF. 276 address. 27 Network Layer. 139–46 destinations for. 193 for SSH. 56 RAM. 104 authentication for. 254 2865. 207 Memory. 103 EIGRP. 33. 204 interior gateway. 16 addresses. 63 release numbering. 265 directed broadcast. 243 Message severity levels. 18 IPSec. 58 RFC. 270 Legal notice. 58 null. 63 G Group SNMP. 207 of passwords. 62 encrypting. 40 route advertisements. 58 CBAC and. 15 NVRAM. 277 interface. 160 FLASH. 103.Index E EGP. 140 IDS Post Office. See Access Control List Firewall. 154 GTSM. 11 switches. See AAA Key chains. 7. 281 virtual. 205 for OSPF. See Memory Remote Administration. 151 viewing logs. 105 1700. 122 Version 1. 61 Protocols by port number. 145 timestamps in.323. 97. 279 rules for. 232–40 FTP. 238 IPSec and. 57 Firewall. 18 Physical Security. 126 Privileges. 166 R RADIUS. 88. 124 Physical Layer. 194 H H. 125. 64 SSH for. 103 IKE. See IPSec Interfaces ACLs on. 73 over SSL. 281 command modes. 57 timeout. 46. 141 SNMP trap logging. 192 3031. IOS and. 115 NTP authentication. 68. 24. 104 authentication. 229. 67 dial-in. 146 Syslog host configuration. 74 L LAN. 268 for administration. 19 Ethernet. 124 826. 77 RADIUS. 88. 213 IPv6. 59 Logging. 217 testing. 283. 24 updating. 232 Network Layer. See CBAC message format. 7. 283 role of. 144. 212 loopback. 44. 145 Syslog logging. 129 passive (routing). 109 for IKE. 204 for IPSec. 108 for RADIUS. 265 Icons. 284 versions. 75 Ethernet. 158. 20 MAC address. 85 binding. 194 key management. 18 address. 104 distribute lists. 63. 235 disabling. 265 IS-IS. 275 viewing status. 111. 123 IKE. 49 and RIP. See SNMP Modes IOS modes. 88–93. 55 Prefix lists. 103 mop. 51 1757. 168 SMTP. 243 M MD5. 76. 39 exterior gateway. 109. 17 1027. 27 NTP authentication. 227 loopback address. 76 redirects. 241–46 Director. 150 configuring. 15. 112 Encryption. 190 IPSec for. 180 and switches. 267 3682. 17 Layers. 104 authentication for. 123 timers. 142 IOS AAA and versions. 11 IGP. 77 IP Addresses. 142. disabling. 235 setting username. 228 for TACACS+. 285–88 viewing processes. 234 HTTP. 88 2267. 161 K Kerberos. 267 N NAT. 279. See AAA RAM. 51 ICMP. See Keys Keys. 118 Filters. 141 MIB. 121 strategies. 15 icon for. 105 Intrusion Detection (IDS). 59 O OSI Model. 91.1c 303 . 150 P Passwords. 106 migration to. 24 EXEC mode. 192 routing protocols. 43 CET. 5. 170 MPLS. 15 Architecture. 120 unused. 205–17 F Filtering. 264 I IANA. 235 TACACS+. 38. 102. 169 Q Quality of service. 71. 160. 76 source routing. 79. 38 dynamic. 149 disabling. 160. See IOS RIP. 256. 156 1918. 19 compare to IPv6. 253. 28.

59 VTY line. 35. 76 disabling. 152 vulnerability in. 219 Kerberos and. 76. 72 syslog port. 68 testing. 267 U UDP. 255 for NTP. 27 Turbo ACLs. 251 Users. 169 port numbers. 69 Security for the Simple Network Management Protocol (SNMP). 60 ACLs on. See RMON v3 security levels. 70. 258 diagnostic commands. 74 CDP. 86 nmap scan. 130 Syslog. 110 Software Updates. 301 304 Version 1. 210. 147 Timeouts for CBAC. 196 SSH clients. 151–55 Security Policy. 7. 300 minicom. 75 TCP small servers. 72 timestamps. 256 SSH clients. 152 Tools CyberCop. 235. 103 null routing. 64 defining local. 270 Unicast RPF. 165 transcript of. 27 and CBAC. 252 automated. 228 RSVP. 278 UDP small servers. 87 log messages on. 126 static routes. 48 Unreachables. 102–30 default. 140. 169 Testing. 116 Checklist. 77 pad. 251 SCP. 87 configuring. 72 V VLANs. 166–73 management of. 72 SYN attack. 47. 39 small servers. 276 viewing status. 227 availability. 143 number of. 86 nmap scan. 146–50 time zone. 161 security concerns. 163 RMON. 70. 43 neighbor authentication. 88. 15 Telnet. 73 syslog. 205 in access lists. 230 S Scanning. 250–57 attack. 75 SNMP. 254 traceroute. 72 directed broadcast. 155–58 Route Table. 250 Transport Layer. 149 open sockets. See IPSec VTY. 130 viewing. 90 RSA. 132 MPLS. 73 NTP. 41 in access lists. 1. 255 functional. 85 intercept. 204 security policy for. 44. 60 protecting. 68 nmap. 228 W Wide-Area Network (WAN). 276 ACLs on. 231 SHA. 106 URPF Verification. 263 VPN. 5. 103 distribute lists. 106 role of. See Logging System and Network Attack Center (SNAC). 46 security testing. 280 Switching. 70. 17 filters for. 77 source routing. 90 keepalives. 230 RAT. 145 using SSH on. 24. 231 TeraTerm. 28. See VTY Services bootp. 190 passwords. 251 NTP port. 16 Windows IPSec and. 6 T TACACS+. 145. 123 version command. 129 passive interfaces. 221 SNMP. 253 configuring. 238 IKE port. 158 procedure for. 251 ntpq. 160 Time. 37 Review of. 16. See AAA TCP and CAR. 119 prefix list filtering. 251 TFTP. 171. 97 and CBAC. 153 RMON. 65 reverse telnet. 167 SSH. 231 diagnosing. 59 for SSH. 238 for EXEC. 39 small servers. 130. 287 clients. 256 SATAN. 59 dial-in. 66 logging port numbers. 73 HTTP.Router Security Configuration Guide migration from. 207.1. 84 logging port numbers. 9. 129. 95 Switches. 92 Routing. 258 SAINT. 250 Router Audit Tool (RAT). 251 port numbers. 46. 233 handshake. 274 finger. 62 viewing logged in. 148 RANCID. 106 unicast RPF and. 103. 168 TCP/IP.1c . 263 CEF. 67. 148 list of. 132 Unix finger protocol. 121 dynamic.