Continuing my expedition in to trying to learn EIGRP and other routing techniques in GNS3, I now have another theoretical network, depicted below:

network diagram

Everything in the red area is a mystery. We do not have access to configure it, we only know how it is connected. We also know that the red area routers only support static routes, and have the corresponding router in the green area set as their gateway of last resort. We cannot change anything in this area.

The blue/lilac area just does layer two switching. Nothing fancy, and cannot be changed.

I'm working in the green area. Both of these routers have EIGRP configured with their directly connected networks, and auto-summary disabled. These two nodes can see each other, and mutually identify as EIGRP neighbours. Thanks to EIGRP, A-R2 can ping B-R2.

The question then, is how can I have these two routers (A-R1 and B-R1) advertise the networks that are behind A-R2 and B-R2 (,, and, so that clients on each of the aforementioned networks can communicate (e.g. PC at can ping a PC at

Any solution should be scalable. I should be able to add another router in the green area, that then handles routing to an additional router in the red area (replicating to C, D, E, F, etc..).

Remember though, that nothing in the red or blue areas can be changed by me.

I have tried simply adding those under the EIGRP network list of A-R1 and A-R2 for the respective router, but that is not the solution here.

Target hardware is Cisco 3660 series routers running C3660-IK9O3S-M version 12.4(13b).

2 Answers 2

  1. There needs to be a few assumptions in place in the red area - A-R2 and B-R2 need to have some kind of upstream static route to A-R1 and B-R1 respectively. This would most likely be a default route.

  2. A-R1 and B-R1 will need static routes pointing back to A-R2 and B-R2 respectively for their specific routes (ex: A-R1 would have routes to and via, B-R1 would route and via

  3. You'd redistribute the static routes from #2 on A-R1 and B-R1 into EIGRP. Note that good practice here is to use a route-map to explicitly match the routes being redistributed and setting a metric on the route.

  • So as an absolute bare-bones, I could configure the static routes (for the respective R2 router's networks) and then use distribute static under my router eigrp x entry on the EIGRP capable routers? That is, of course, ignoring any sensibilities around trust in my configuration and others. A real-world implementation would need to be more robust with use of route-map to filter, as well as appropriate metrics. Jul 13, 2020 at 4:24
  • 1
    The route-map is also where you set the metric for the route. You can also specify a metric for all routes at the point of redistribution in EIGRP (redistribute static metric xyz). Using a route-map and explicitly specifying what's allowed, what's not and what gets what metric is a basic good practice for route redistribution and can save you a lot of future pain - particularly when redistributing between dynamic routing protocols.
    – rnxrx
    Jul 14, 2020 at 1:05

If the R2's aren't running a routing protocol with the R1's, the only way to know what's beyond them is via statics. And that's not exactly "scalable", but since you don't control them, you can't trust what they might advertise anyway. (i.e. even running a dynamic protocol, you'll have to explicitly configure a list of allowed prefixes.)

(Hint: the global internet uses various routing registries to help automate the building of those prefix lists.)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.