6.5 Enhanced IGRP Operation
6.5.3 Choosing routes
EIGRP route selection is perhaps what distinguishes it most from other routing protocols. Following are its key characteristics:
  • EIGRP selects primary and backup routes that are kept in the topology table (up to six per destination). The primary routes are then moved to a routing table.
  • Like OSPF, EIGRP supports several types of routes: internal, external (that is, non-EIGRP), and summary routes.
  • EIGRP uses the same composite metric as IGRP to determine the best path. The metric can be based on five criteria. The default criteria used are as follows:
  • Bandwidth ---The smallest bandwidth between source and destination
  • Delay --- Cumulative interface delay along the path

Following are additional criteria that can be used. These criteria are not recommended for use because they do not result in the path selection choices that you may think.

  • Reliability --- Worst reliability between source and destination based on keepalives
  • Loading --- Worst load on a link between source and destination based on bits per second
  • MTU --- Smallest MTU in path

Note: EIGRP doesn't sent updates unless there is a topology change. The reliability or loading of a link could change dramatically but this wouldn't necessarily change the topology. EIGRP, then, would not send any updates to advise of the change. Therefore, enabling the use of reliability or load does not really provide any additional intelligence to the path selection process.

EIGRP uses the DUAL algorithm to calculate the best route to a destination. DUAL selects routes based on the composite metric and ensures that the selected routes are loop free. In Figure , the EIGRP metric determines that the three-hop path using the T1 lines is a better route that the single 19.2-kbps link. RIP would select the single 19.2-kbps link by selecting the path based on hop count.

EIGRP uses the following process to determine which routes to keep in the topology and route tables:

  1. DUAL is run on the topology table to determine the best and loop-free primary and backup routes to each destination.
  • "Best" is the lowest metric route that is calculated by adding the metric between the next-hop router and the destination (referred to as advertised distance) to the metric between the local router and the next-hop router (the total is referred to as feasible distance). For example, in Figure , from router A, the advertised distance to network 7 using router B is 21, and the feasible distance is 31 because of the additional link cost between routers A and B, which is 10.
  • The next-hop router(s) selected as the best path is referred to as the successor. Multiple successors can exist if they have the same feasible distance. All successors are added to the route table. In Figure , router B is the successor for network 7.
  • The next-hop router(s) for the backup path is referred to as the feasible successor. If the successor route is no longer valid and a suitable feasible successor exists, this feasible successor replaces an invalid successor in the routing table without a recomputation. More than one feasible successor can be kept at one time. These routes do not need to have the same feasible distance, but their advertised distance must be less than the feasible distance of the successor route.
  1. The successors and feasible successors are kept in the topology table, along with all other routes, and referred to as possible successors. The only routes removed are those that have a metric of infinity (unreachable).

Note: EIGRP uses the same composite metric as IGRP to determine the best path. The default criteria used for that metric are bandwidth (the smallest bandwidth between source and destination) and delay (cumulative interface delay along the path).