[c-nsp] routing between VRF and global

[Pavel Skovajsa]

Hello Jeff,
Yes, sound strange, but everybody does this.
>From my experience it seems like the only purpose to split the network into
VRFs is to subsequently join these VRF due to various business requirements
:)

I learned most of the stuff from the MPLS Architectures Volume 2 book. Their
solution is to inject the routes into MP-BPG and import them in your VRF
config. If you search the archives you may be able to find some examples as
well.

-pavel skovajsa

On Fri, Jul 16, 2010 at 3:17 PM, Jeff Bacon <bacon at walleyesoftware.com>wrote:

> I have a mesh of 6500s connected via various gig fiber links. The 6500s
> have multiple VRFs defined, but of course most things interesting live
> in the global zone.
>
> I want a host on a VRF on a 6500 to be able to connect to another
> destination that is reachable through the global zone. Most likely it
> will be on the same 6500, but ideally it would be the same one way or
> the other.
>
> Basically, how do you leak routes between VRF and global? Between VRF
> and VRF I get. VRF<>global, not so clear; "MPLS fundamentals" provides a
> couple of examples but it's aimed more at a "how to connect VRF to
> internet so you have one static global route entry... ick.
>
> I can see the possible solution of creating a GRE tunnel within the
> switch itself, with one end in the VRF and the other end in the global
> and using "tun vrf" to get them to link, but this seems just a shade
> ugly (though it also happens to provide a nice fixed point in space for
> applying ACLs, etc.)
>
> Or of course there's the "hairpin" solution. I might be able to live
> with that, probably better than the GRE answer... but that doesn't mean
> I have to like it, does it? :)
>
> Thanks,
> -bacon
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>