Border Node Redundancy

[dm2020]

I have the attached external border topology. For the underlay network we have a link between each border node and each fusion switch and a single link between border nodes. There is an eBGP adjacency over each of the border to fusion links (4 in total) and an IGP adjacency between borders for the underlay

What I want to confirm is the need for East-West iBGP peering between the border nodes for each VN if we have adequate North-South eBGP redundancy. Looking at the diagram above, there is an eBGP peering over each link between the border nodes and fusion switches which provides continuous forwarding in the event of a single external link failure. This is automated by DNA Center for each VN. From my understanding, iBGP between borders will only be used in the event that one border loses all of its external links which is an unlikely scenario. DNAC doesn't automate the iBGP peering configuration so I want to avoid the need to manually configure if it’s simply not needed. Is there any reason to configure iBGP in this scenario?

Thanks

[dm2020]

Sorry, duplicate post

[Scott Hodgdon · Accepted solution]

dm1981,

It is not a requirement for East-West iBGP between borders, but it is a Best Practice for maximum resiliency. If you feel that you have sufficient resilience northbound of the Border, then the East-West is not needed. That said, you will most likely have an East-West link between borders for underlay resilience as well, and often this is a trunk that carries both iBGP and IGP.

Cheers,
Scott Hodgdon

Senior Technical Marketing Engineer

Enterprise Networking Group