EIGRP Stub - It is actually one of the EIGRP Scalability features but also it helps many other things in EIGRP. Also, in this post, we will share a topology that will be used to explain some design caveats with EIGRP network design. Before we explain the EIGRP Stub, let me explain some EIGRP convergence behaviors.
EIGRP ConvergenceIf you are looking for much more detail on EIGRP Design and Practical Labs, have a look at our EIGRP Training.
When the EIGRP node loses the Connection to the prefixes. If there is no feasible successor installed in the EIGRP topology database. The router is marked as active and the EIGRP query is sent to every neighbor. In the above topology, Router D doesn’t know the 192.168.0.0/24 network. Router C sends a summary 192.168.0.0/16. That’s why it replies without asking Router E. Router B has an alternate path, thus, Router B replies immediately. Router J doesn’t have any EIGRP neighbors.
It replies to the Query immediately. Router G doesn’t know the 192.168.0.0/24 network. Router F filters the 192.168.0.0/24. That’s why Router G replies without asking Router H. So, as you can see, even if you filter or summarize the prefixes, EIGRP Query is sent to the neighbor and the neighbor also send a query to the router that is one more hop away.
EIGRP Stub FeatureIf you want to stop EIGRP Query to be sent completely, you can only do it EIGRP Stub feature. With the EIGRP Stub feature basically, you are creating an artificial split horizon for prefix advertisement!. Wow, this is so nerdy sentence, let me explain with the below topology :)
In the above figure, if one of the routers in the spoke site 1 receives prefixes from the hub router, it doesn't advertise to another spoke on the same site, although they are on the same site if the EIGRP Stub feature is enabled on those spoke routers.So, with the EIGRP Stub feature, when a router receives a routing advertisement, it doesn't advertise to another EIGRP router. Classical Split Horizon. Same mechanisms we have in IBGP as well. Thus, in the above figure, when there is a failure, spoke routers can be isolated from the rest of the network.
Either, you should connect the spokes to both of the Hub routers, or you need to leak the prefixes between the spokes, or you need to put a static route for the rest of the network reachability on the spoke routers with higher AD. This design caveat will be explained in another post in more detail. If you are looking for more EIGRP posts, please refer to our EIGRP category.