If I have a simple topology with three routers,

             R2  -- stub area 1
ISP ----- R1
             R3  -- stub area 2

and I want to alleviate both R2 and R3 from being single points of failure for their stub areas, but (for route summarization purposes) I do not want to connect stub area 1 to R3 and vice versa, could I simply add a fourth router, connect each stub area to that router and that router back to R1, and run GLBP separately on each interface for the stub area to which it is connected, like this:

         R2  --- Stub area 1
        /         /
       /         / (fa0/0)
ISP -- R1  ---- R4
       \         \ (fa0/1)
        \         \
         R3  ---  Stub area 2

and configure fa0/0 on R4 to be in a GLBP group with R2, and fa0/1 to be in a GLBP group with R3?

I don't see any reason why this wouldn't work, but for some reason all of the illustrations I see of a GLBP topology always involve a separate router for each GLBP group.

  • What routing protocol are you using? How many routers are there in each of these areas? OSPF would be a much better solution, there shouldn't be a problem advertising both summaries from two places. In your diagram R1 is a single point of failure, so you really need to look at providing resilience there. Apr 11 '18 at 5:54
  • @KarlBillington No routers in the stub areas, just hosts. Protocol is OSPF for now, but that's not set in stone. What do you mean advertising both summaries from two places? You mean to just keep the original topology and connect each stub area to the other router, and advertise both summaries on both routers? || Yes, R1 is a single point of failure also, but one thing at a time... Apr 11 '18 at 17:21
  • OK, you could just connect both LANs to R2 and R3 and advertise both networks from both R2 and R3. To answer your original question, yes, you can have GLBP operate on two different interfaces at the same time, each one will be a different instance Apr 11 '18 at 17:55

Yes, it is possible to run two separate instances of GLBP on different interfaces on the same router. A core/distribution switch may have many GLBP groups configured when serving a large campus.

It may be a simpler solution to remain with the first topology and connect both networks back to both R2 and R3. Both routers can then advertise both networks towards the ISP. If you do connect both networks to both routers, make sure there is resilience in the LAN so that you don't end up with a partitioned LAN during a link failure. That would lead to inbound traffic not reaching some hosts. Also, consider whether you need to load balance at all. If you used VRRP instead, it is an open standard and would give you more vendor options in the future.

At the end of the day, this helps with single point of failure to the LAN, but there are also single points of failure with R1 and the ISP connection.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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