vrf aware mpls LxVPN's ?

I'm faced with the following..

Network 1 - I have a network that's been around for a few years or more.All
IP/Ethernet, no mpls, no vrf

Network 2 - I now have a (10) node asr9k ring with asr901's on the edge,
fully mpls enabled, and already providing a few l2vpn's..with plans for much
more

At the core of network 1 are dual 7609's

I have colo'd with those 7609's an asr9k also. We've made a 10 gig
connection between the 7609 and 9k at each site of the 2 core sites.

I wanted to extend the mpls logic into the 7609. HOWEVER, I wanted the core
ospf/mpls tables within my new 9k cloud to NOT comingle with the tables in
the 7609's (public route tables).perhaps this is a typical design goal of
many of you out there doing similar things.

So I thought I would edge mpls into the 7609's using a vrf on the 7609's.
but I was told that the vrf's in the 7609's will not support mpls LxVPN's
*within the vrf* ..this approach seems to be opposite of how you would
normally do it (in other words, typically, as I understand it, it's mpls
that's global and it's the ip/L2 services that are put into vrf's)..but as
you can see I'm faced with HOW-TO migrate my (2) existing 7609's into an
mpls network as PE's.

Someone suggested to me to move all 7609 routing currently in existence into
a vrf..then interconnect the 7609 and 9k networks so that the default
vrf/global routing in the 7609 is now the SP core tables I wanted to keep
isolated from all the LxVPN overlays. that seems major as I sit here
pondering that. I think I might, in the short-term, land the mpls
(pw/vpls/etc) stuff on the asr9k colo'd with the 7609 and do flow points via
traditional 1q trunk to 7609....any problems with that ? ....then down
the road perhaps overhaul the 7609 as previously stated.....however, by then
I may have moved a bunch of stuff physically off of the 7609 and onto that
colo'd 9k, which would make it moot

Ideas/thoughts?

Aaron

Blogs and organic groups at http://www.ccie.net
Received on Wed Apr 11 2012 - 11:23:53 ART