Configuring IGRP

Technical Overview of RIP

RFC 1058 outlines RIP-1, and RFCs 1721, 1722, and 1723 are supplemental RFCs that allow for RIP-2 extensions.

NOTE

You can find all RFCs online at http://www.isi.edu/in-notes/rfcxxxx.txt, where xxxx is the number of the RFC.

RIP operates on UDP port 520 ”that is, all RIP packets have a source and destination port equal to 520. RIP operates in the following manner:

1. Initialization ” When the RIP process initializes, it sends a request packet out each participating interface. This request packet asks for a full routing table update from all RIP routers. This request is accomplished by broadcasts on LANs and is sent to the next -hop address on point-to-point links. The request packet is a special request, querying the neighboring devices for a full routing table update.

2. Request received ” RIP has two message types, a response and a receive message. Each route entry in the request packet is processed and a metric is formed for the route along with the path to the route. RIP uses the hop-count metric. A hop count of 1 means a directly connected network, whereas a hop count of 16 denotes an unreachable network. The entire routing table is sent back as a response received message.

3. Response received ” The router receives and processes the response. It can update its routing table with the addition, deletion, or modification of entries.

4. Regular routing updates and timers ” Every 30 seconds, the router sends out the entire routing table in the form of a response message to every neighboring router. The router sets an expiration timer or invalid timer of 180 seconds when a new route is received or an update about an existing route is received. If no new update for a route is received within 180 seconds, the hop count for the route is set to 16 (unreachable). The router advertises the route with a metric of 16 until the flush timer removes the route from the routing table. The flush timer is set to 240 seconds, or 60 seconds longer then the expiration timer. Cisco's implementation of RIP, which is not defined in RFC 1058, uses a third timer called a hold-down timer. The hold-down period occurs for 180 seconds after a route is received with a higher metric. During this time, the router does not update its routing table with the new information that it received, to provide additional time for convergence.

5. Triggered routing updates ” Whenever a metric for a route changes, the router sends only the routes that are affected by the metric change. A full routing table is not sent.

All the timers can be changed with the timers basic update invalid holddown flush command.

CAUTION

When adjusting the timers for RIP, all the timers in the routing domain must also be adjusted. Failure to do so can cause many unpredictable results.

Classful Routing (RIP-1 Only)

RIP-1 is a classful routing protocol, so it does not advertise a subnet mask along with advertised routes. For RIP to determine what the subnet mask is of the destination network, RIP uses the subnet mask of the interface in which the route was received. This is true only if the route received is a member of a directly connected major network. If the route received is not of the same major network, the router tries to match only the major bit boundary of the route ”either Class A, B, or C. For this reason, it is critical to preserve a consistent bit mask in each major network throughout the entire RIP routing domain.

Figure 9-1 illustrates classful routing on a RIP-1 network. Let's examine how the wolverine and rogue routers handle the RIP process.

Figure 9-1. Classful Routing Example

The wolverine router has two interfaces in the 128.200.0.0 major network. The 24-bit mask is consistent on every interface on the router. Therefore, the router receives updates only for the major network 128.200.0.0 on the 24-bit boundary. When the router receives the update out of this major network ”for example, 192.16.1.4/30 ”it enters only a summary route in its routing table pointing to the major bit or class boundary of the address, summarizing the address to 192.16.1.0/24. Example 9-1 shows the output of debug ip rip demonstrating this concept.

Example 9-1 debug ip rip Command Output on wolverine

 01:24:39: RIP: sending v1 update to 255.255.255.255 via Ethernet1 (128.200.1.1) 01:24:39:      subnet  128.200.2.0, metric 1 01:24:39: RIP: Update contains 1 routes 01:24:39: RIP: Update queued 01:24:39: RIP: Update sent via Ethernet1  01:24:39: RIP: sending v1 update to 255.255.255.255 via TokenRing0 (128.200.2.1)   01:24:39:      subnet  128.200.1.0, metric 1   01:24:39:      network 192.16.1.0, metric 2  01:24:39: RIP: Update contains 2 routes 01:24:39: RIP: Update queued 01:24:39: RIP: Update sent via TokenRing0 

The rogue router has two interfaces. One is in the 128.200.0.0/16 major network, and the other is in the 192.16.1.0/24 network. When the rogue router receives the 128.200.2.0/24 and 128.200.1.0/24 subnets from its E0/0 interface, it tries to send them out its E0/1 interface. Because this interface has a different mask (30-bit), only the summary route 128.200.0.0/16 is sent to the cyclops router. Example 9-2 demonstrates this happening, through the aid of debug ip rip.

Example 9-2 debug ip rip Command Output on rogue

  RIP: received v1 update from 128.200.1.1 on Ethernet0/0   128.200.2.0 in 1 hops  RIP: sending v1 update to 255.255.255.255 via Ethernet0/0 (128.200.1.2)      network 192.16.1.0, metric 1  RIP: sending v1 update to 255.255.255.255 via Ethernet0/1 (192.16.1.5)   network 128.200.0.0, metric 1  

Classless Routing (RIP-2 Only)

RIP-2 is not a new routing protocol, but it is an extension of RIP-1 provided by RFCs 1721, 1722, and 1723. The extensions provide the following enhancements to RIP:

• VLSM support. The router carries the subnet mask in the update, which allows the router to handle VLSM addressing.

• A next-hop address carried with each route entry.

• Support for external route tags.

• Multicast route updates.

• Support for MD5 authentication.

The most significant of all the enhancements is the support for VLSM, making RIP-2 a classless routing protocol. It is no longer critical to have one bit mask throughout the entire routing domain.

Most of RIP-2 operational procedures and timers are identical to those of RIP-1. RIP-2 uses the multicast address of 224.0.0.9 to send updates versus the general all- hosts broadcast used by RIP-1.

RIP-2 is fully backward compatible with RIP-1. This is accomplished by the means of a compatibility switch and a receive control switch, as defined in RFC 1723. Essentially, these switches allow you to control what type of RIP updates the router sends and receives. The router can be configured to receive only Version 1 updates, only Version 2 updates, both, or none. The router can send only Version 1 updates, send Version 2 updates as a broadcast message, send Version 2 updates as a multicast, or send no updates. The switches can be manually set with the following interface command:

  ip rip  [  send   receive  ]  version  [  1   2   1   2  ]