To speed things up and to take the ease of OSPF we are exploring to migrate the "edge" networks to BGP with recursive routing (OSPF+BFD+MPLS+BGP) We have done some laboration with a a couple of CHRs and it works great, using the "hub"-routers with route-reflection.
Any one running this kind of setup or having some suggestions/issues with it?
R1 ----> R2 -----> R3
If I understand correctly:
- The routes gets installed (R1) by BGP with the foreign Router-ID (nexthop self) in the system table. This is not direct routable via L3 but the MPLS layer has an label for that Router-ID, thus it gets forwarded out with that label.
- On the neighbouring routers (R2) it has the same Router-ID in its MPLS forwarding table and just swaps the label for the appropriate one.
- When it arrives at the target router (R3) with that Router-ID, it drops the label, and then of course it has that network in it's local routing table.