Redistribution from BGP to LISP

[iores]

Hi,

In SDA, what is the reason of redistributing BGP routes to LISP at border/control node?

I mean, with Proxy-xTR, every edge node will forward packet to it if it receives negative map reply from MS/MR. Then, Proxy-xTR will perform another check with MS/MR and receive negative map reply with indication that packet should be forwarded natively.

What am I missing?

[Preston Chilcote · Accepted solution]

The Cisco Live session BRKENS-2502a goes into some detail of this at the 28:30 mark.  It sounds like there was no other way to get external routes from Border to non-colocated Control plane node until LISP was enhanced. 

For new SDA deployments LISP pub-sub is recommended and BGP isn't needed inside the fabric.

[jalejand · Accepted solution]

Redistribution of LISP Into BGP is to announce an aggregate fabric subnet to upstream networks

BGP route-import into LISP (route-import database bgp xxx etc) is a feature intented for internal borders to allow being the exit point of a particular prefix.

In a topology where both Borders are mirrored and they expect to be the exit point in the form of load balancing, importing BGP routes into LISP is not necessarily a good thing, in fact is just additional control plane operation that results in the same thing: equal path routing to each border. In fact, external-only borders for these type of topologies is a better choice.

But if you want to force traffic for a specific prefix to go out of a particular border, you can take benefit of importing a BGP route into LISP, so its preferred over the external borders which do not import this route.