[c-nsp] VRF question

dave bernardi dave.bernardi at adelphia.net
Mon Oct 11 16:41:55 EDT 2004

At 04:45 PM 10/11/2004 +0200, Per Carlson wrote:
>On 2004-10-11 16:25, Rodney Dunn wrote:
>>W2N-3.6-7401ASR-B#sh run int gig 0/0
>>Building configuration...
>>Current configuration : 146 bytes
>>interface GigabitEthernet0/0
>>  ip address
>>  tag-switching ip
>>The interface looks just like any other MPLS interface.
>>Just globally you would prevent labels for being exchanged
>>via 'tag-switching advertise-tags <acl>'.  What happens
>>in the forwarding path depends on the control plane being
>>set up and tags exchanged for the relevant prefixes.
>>In this scenario you can have ip and mpls packets going
>>in/out the same interface.  
>Looks like I have to brush up my MPLS knowledge ;-) I my world of MPLS (L3 VPNs), either packets are tagged (MPLS side) or untagged (VRF side) on a P/PE. Appearently, you could do a lot of other nifty things with MPLS which I don't grap (yet).

 From my brief observation in the lab..... Set up one P device with 3 edge devices (2 as LSRs and 1 as normal IP router). With LDP enabled on the P and PE devices all traffic between them is MPLS tag switched (LDP neighbor relationships have been established). Any traffic from any PE router to the IP router is untagged because no LSP can be created or tags exchanged since MPLS is not enable on the IP router.  show mpls forwarding-table shows the distinction between tagged and untagged.

So from that brief test it was my understanding that the interface sub command "tag-switching ip" doesn't limit the interface to labeled packets only but rather allows processing of tagged and untagged packets. This is a good thing otherwise enabling MPLS in a production network would be quite tricky.

>I think it's time to end this branch of the tread now.
>cisco-nsp mailing list  cisco-nsp at puck.nether.net
>archive at http://puck.nether.net/pipermail/cisco-nsp/

More information about the cisco-nsp mailing list