Simple TE question

[djankovic]

Hello. Let's say I configure explicit TE path betwen two edge devices and it goes on suboptimal route. How should I then do a configuration so only traffic from L3 VPNs that I select goes through TE tunnel, and all other traffic goes through normal path with smallest IGP metric.

[swaroop.potdar]

Create additional loopbacks per vrf ( in global table) on either side and manipulate the BGP nexthop to point to that new loopback, instead of regular loopback 0.

For eg:

Create loopback 1 in global for vrf1 on the egress PE. On the Ingress use a global static loopback 1 route pointing to the tunnel which terminates on this egress PE.

Manipulate the nex-hop in the MPBGP update on the ingress using import map pointing to the loopback 1 on the egress.

Or

Assign the BGP next-hop to be used for that VRF as loopback 1 in vrf config mode.

HTH-Cheers,

Swaroop

[Harold Ritter]

Swaroop,

Or even easier. If the goal is to follow the tunnel for all l3vpn traffic, just use a separate loopback interface address for the VPNv4 session and make that loopback address reachable via the tunnel interface using a static route. This way you don't need to se a separate loopback per VRF.

Regards,

[djankovic]

No, actually the goal is to tunnel traffic only for certain VRFs. Most of VRFs should still pass regular path.

[Harold Ritter]

Setting the bgp next hop under the given VRFs seems like the best option then:

ip vrf test

bgp next-hop loopback 1

Regards,

[dara111983]

Hi Harold,

Can U suggest us any link wherein this solution has been documented???

This is a very nice flexible way of giving the service using specific TE tunnel to high priority customers requesting optimization within SP core.

[swaroop.potdar]

Here is the command reference for this toggle.

Although its not documented as a feature, so doesnt have a feature guide as far as i know.

http://www.cisco.com/en/US/products/ps6441/products_command_reference_chapter09186a00804a3736.html#wp1022044

If you want further granularity based on subnets, (for eg: only intra VPN traffic is High priority and traffic to internet is not etc.) then you may want to manually modify the next-hop rtaher than assigning the next hop per VRF using the first method which i specified in my earlier post, if you are looking for something like a Virtual leased circuit kind of service between end points then you can gurantee all traffic certain SLA using the second method 'the per VRF next-hop method'.

HTH-Cheers,

Swaroop

[dara111983]

Swaroop's response at lightning speed :) :) :),

Thanks,

Regards,

Dara