I have a setup where we maintain our own MPLS core with multiple VRFs. We are moving to a next phase where we want to do inter-vrf routing. At a high level, if I have the following:
VRF A - path to internet, (default route is available), path to tools network
VRF B - Business Group B
VRF C - Business Group C
VRF D - Business Group D
If we do something where we want VRF's B, C and D to receive the default route from A. That also implies that A has to have the networks from B, C and D in its routing table to send traffic back to. VRFs B, C and D do need a path to the internet via VRF A.
With this setup, we don't want B, C or D to be able to talk with each other.
Scenario:
If we have the default route from A leaked into the VRFs B, C and D and VRF A has the routes from the other VRF's.
If a Device in VRF B sends traffic to an IP address that is actually in VRF C. It would use the default route it has to get to VRF A. Since VRF A has all the routes, it would have the more specific subnet of the destination address for the subnet in VRF C and would be able to send the traffic to VRF C. On the return path, VRF C would use the default route to get back to VRF A and VRF would then have the more specific subnet to get back to VRF B.
does the above sound like something that could happen at a high level or is my understanding of what might happen wrong?
Is there a good way we can avoid having VRF A be a transit network between the other VRFs when using the default route?
As mentioned, VRF's B, C and D do need the default route.
