At what stage is the RT applied to the prefix?

vdadlaney · ‎10-24-2008

I am running a vrf aware IBGP session between 2 neighbors. This is across a Layer 2 trunk interface that is part of the VRF. The behavior I am seeing is that the RT does not get propagated to the neighbor. Infact the RT that gets applied to the route is the one that is locally configured as the export RT. If there is no export RT configured than there is no RT applied to that route. Is this the correct behavior?

Usually when we have a CE-PE type scenario the RT gets applied once the route enters the PE. It seems correct that the RT in a CE-CE type instance would work the same however on the same box when we have 2 VRF's configured and a route gets propogated from 1 VRF to the other VRF the route still has the RT attached from the 1st VRF. Thx

Giuseppe Larosa · ‎10-25-2008

Hello Vikram,

RT is a BGP attribute of type extended community that is attached to a VPNv4 prefix in address-family vpnv4.

the vpnv4 prefix is made of:

Route_Distingusher:IPv4 prefix

the first is 64bits wide.

even if in simple scenarios RD=RT as value they are different concepts in MPLS VPN.

RD is there to provide support for address overlapping and segregation within a single PE.

RT is used to export routes to the core and as the criteria to decide to import routes in a VRF.

RD is unique

RT can be multiple values as happens for BGP community.

So if you mean a VPNv4 iBGP session if you don't set the export RT in the VRF config your routes will be without any RT value and shouldn't be imported in a VRF in the second PE.

But if it is an iBGP session within a VRF in address-family ipv4 vrf VRFname

the RT concept never applies to it because they just exchange IPv4 prefixes like with a normal CE.

Each sees the peer as a CE only router.

So what you see is correct.

Hope to help

Giuseppe