Enhanced Interior Gateway Routing Protocol (EIGRP) is considered deprecated routing protocol. In my opinion unfairly. Despite EIGRP is distance vector protocol (or rather hybrid protocol) is still being utilized in small and medium networks. Reliability and efficiency in connection with simplicity of deployment make EIGRP still worth of considering routing protocol, especially, that has DUAL algorithm .
EIGRP is Cisco proprietary, but has been released and converted into open standard in 2013 what means that from then on, can be implemented on other vendors devices. It belongs to distance-vector protocols group, what means that doesn’t have knowledge about a whole network topology as link-state protocols. EIGRP keeps 3 tables in memory : neighbor table (which contains the list of routers and its interfaces directly connected to) topology table (which contains the list of routes learned from neighbours) and routing tables with best paths to destinations. What distinguishes EIGRP from other distance vector protocols is that between neighbors are sent updates only, not entire routing tables. This is why EIGRP is being called hybrid routing protocol. It has its own protocol number 88 and uses HELLO packets exactly like OSPF to establish neighbourhood and wide set of metrics : bandwidth, delay, load and reliability.
The heart of EIGRP is DUAL (Diffusing Update Algorithm) which bases on 3 tables (mentioned above) calculates the SUCCESSOR and FEASIBLE SUCCESSORS. The Successor is selected by calculation the most efficient path. In order to do that EIGRP uses Advertised Distance (how far the destination is away from its neighbor) and Feasible Distance (the total distance to the destination from the edge port from the router point of view). The Feasible Successor is nothing more than backup path (or paths). Having backup path significantly increases speed of convergence. In case of failure EIGRP doesn’t calculate new path but uses backup path immediately How is being selected ? The path can become a backup path if has a lower advertised distance than total distance (feasible distance) of your best path.
Advertised Distance of Feasible Successor < Feasible Distance of Successor
Let’s lab it out!
Simple topology, the numbers on the link represent the ‘administrative distances’ in black and ‘feasible distances’ in red, according to the rule that says that : if Advertised Distance is lower than Feasible Distance of the Successor then the path becomes the Feasible Successor. And this is what we exactly see, 20794 is lower than 28160 and path through the Cork becomes FS, Dublin path is Successor. We focus on ‘Source’ router, we want to reach ‘Warsaw’ router with network 188.8.131.52
Also betwen particular routers are configured networks:
Source – Limerick : 192.168.1.0
Source – Dublin : 10.0.1.0
Source – Cork : 172.16.1.0
Limerick – Destination : 192.168.2.0
Dublin – Destination : 10.0.2.0
Cork- Destination : 172.16.2.0
Firstly let’s have a look on the routing table. We may conclude the best path to Warsaw router leads via Dublin router.
Now, let’s check topology table. We see to Warsaw router (184.108.40.206 network) we have Successor via 10.0.1.2 (Dublin) with Feasible Distance 30720 and Feasible Successor 172.16.1.2 (Cork) with Advertised Distance 30464.
If we have 2 paths that maight be used, why we don’t use them both actively ? We may use Feasible Successor if we utilize ‘variance’ command. We have to use ‘variance’ command with appropriate multiplier that increases the value of Advertised Distance of Feasible successor. Now, depending on the multiplier and the final value that we got we have different unequal load balancing. In our case from now on, more traffic will be going through the Cork router.