[c-nsp] vpnv4 vrf EIGRP confusion

[Andrew Miehs]

Dear all,

I am a little confused as to how VRFs, vpnv4 and EIGRP work together.

I have the situation that two PE routers (6500s) are connected via MPLS (via the core). These two PEs also have a second direct connection (ethernet) between themselves.

I has hoping to use this backdoor link for one of the VRFs, so I thought it should be possible to run EIGRP over this link.

I have read about bgp cost community, but I do not quite understand how the routing table is organized in the individual VRFs.
I have read that administrative distance is not checked first, but do not understand how this really works.
I have managed to get this to work, but ONLY if I redistribute EIGRP/BGP on both PEs and then "bgp bestpath cost-community ignore".

I had assumed that each VRF has its own routing table, and that this is populated from the routing tables in the various protocols.
Is the VRF routing table really the BGP one?

Does anyone have a link to a technical description of how the VRF tables are actually implemented?

Confused.

Regards

Andrew