MPLS TE question - multiple tunnels between same PEs

[gregcox]

My example is I have 2 PEs, with multiple paths, I have 4 TE tunnels defined between the same 2 PEs, how can I force 4 individual vrf's to take one of each of the TE tunnels.  When all the vrf's are coming in on the same physical interface(via subinterfaces).

From an EXP bit point of few, all the vrf's use the same EXP bits so Class-Based isn't an option.

Both PEs are 6500 Sup720-3B, running SXH.

Thanks for any help,

Greg

[Laurent Aubert · Accepted solution]

Hi Greg,

What you can do is to define 3x loopback's in addition to the one you are already using for MP-iBGP peering. Then configure the PE to use a different loopback as the next-hop for each vrf:

ip vrf 1

bgp next-hop loopbackx

!

Then create a static route on the other PE so he will join this next-hop via a TE tunnel

HTH

Laurent.

[gregcox]

I've tested that, and it seems like it would work from a BGP point of view, but the OSPF/RSVP Signalling doesn't seem to work to anything other than the MPLS TE Router-id loX interfaces, even if the exact same path known good signaling end to end.

Unless I'm missing something on the extra loopbacks, to allow them to be used as MPLS TE destinations, other than making sure they are in the same Area.  As soon as I make the extra loopback the mpls te router-id the extra tunnel comes up, and vice-versa(back to the original), same known good path between.

[Laurent Aubert · Accepted solution]

The idea is not to use those loopback's as TE destination addresses but as BGP next-hop only and to resolve them via your TE tunnels.

So create your TE tunnels as usual, each one with a different explicit-path.

HTH

Laurent.

[gregcox]

oh, that might be it, I see what you mean, missed that first, will test.

Thank you so much.

[gregcox]

Worked perfectly, thank you again for the help.  I would assume the same would be possible with xconnects, do you know if that is the case?

[Laurent Aubert]

With xconnect, you can specify the TE tunnel via a pw-group.

Thanks,

Laurent.

[sivam siva]

Hi @gregcox  @Laurent Aubert 

Even if we have different next-hop for each VRFs, still using the same destination for all 3 tunnels right,

Can you please help to understand how these different nex-hops are mapped to particular TE tunels?