OSPF router (Cisco 3560 L3 E image) has two default routes

[juan-ruiz]

The Cisco layer 3 switch has two default routes with same distance and same metric.

It also has two OSPF neighbors. All three devices that participate on this Vlan6 network belong to a Metro E cloud.

I would like to learn both paths of the default route but only have one in the routing table.

What would be the best approach to have this L3 switch have 172.16.9.2 as the primary in the routing table and 172.16.9.1 as the backup in the OSPF database?

Thanks a bunch for all that have looked at my question.

O*E2 0.0.0.0/0 [110/1] via 172.16.9.2, 04:52:16, Vlan6

[110/1] via 172.16.9.1, 04:52:16, Vlan6

Neighbor ID Pri State Dead Time Address Interface

192.168.200.252 1 FULL/BDR 00:00:39 172.16.9.2 Vlan6

192.168.200.253 255 FULL/DR 00:00:37 172.16.9.1 Vlan6

[lgijssel]

If the paths are really equivalent, the preferred path could also be via 172.16.9.1 and have .2 as backup yes?

Load balancing in ospf is on by default over a maximum of four paths.

Use the maximum-paths 1 command under OSPF to allow only one path to the routing table.

OSPF allows only equal cost load balancing. The link below discusses the topic a bit more extensive:

http://www.cisco.com/en/US/tech/tk827/tk831/technologies_tech_note09186a0080094806.shtml#backinfo

The alternative would be to change the metric of one of the default routes but this may not be part of your administrative domain.

regards,

Leo

[juan-ruiz]

Hi Leo,

Thanks for the fast reply.

Ok so by setting the max paths to 1 I will have one route active in the OSPF and if that link were to be down then the other one would surface in the routing table?

How would I change the metric on the layer 3 switch for the incoming route from 172.16.3.1 to be higher than 172.16.3.2 without making the change itself on 172.16.3.1?

Thanks,

Juan

[lgijssel · Accepted solution]

Ok so by setting the max paths to 1 I will have one route active in the OSPF and if that link were to be down then the other one would surface in the routing table?

Yes, that is the idea.

How would I change the metric on the layer 3 switch for the incoming route from 172.16.3.1 to be higher than 172.16.3.2 without making the change itself on 172.16.3.1?

This is not really possible. You can only manipulate the metric where it is advertized and, as I said already, you may not be the admin of the routers advertizing these default routes. In that case, it will still work as long as you have no absolute preference as to which path to prefer. OSPF will then sort this out for you and the redundancy will work as desired. After a failure on the first path, the alternate path will remain in use until that fails and so on.

While observing the metric of 110/1, it may be that the routes are originated by two ISP routers redistributing them from BGP for example. Please check with your provider that the routes are actually removed when there is an ISP failure. Otherwise you may think you have created a redundant situation where in fact you end up blackholing traffic because the default route is not revoked in case of a failure in the ISP network.

regards,

Leo

[juan-ruiz]

Leo,

I made the change and it worked the max paths under the OSPF config

I do manage both of the upstream routers and I can control them so I could advertise at the router level the metric but I don't want to change anything in that environment.

I was wondering if some kind of distribute list applied to the neighbor would allow the change of admin distance per neighbor or metric per neighbor.

[lgijssel · Accepted solution]

I was wondering if some kind of distribute list applied to the neighbor would allow the change of admin distance per neighbor or metric per neighbor.

It depends on how the metric is achieved.

You can have used for either redistribute or default-information originate.

Each of them requires a slightly different approach.