Tie Breakers with OSPF E2 routes

grichardson661 · ‎10-26-2010

I put this post (below) on www.networking-forum.com and have been researching it this afternoon but not yet found an answer.

Anyone got any ideas.

Hi,

This has got me stumbled. I have a pretty straight forward lab. 4 Routers as follows.

Router A - ASBR in area 0, redistributing EIGRP routes into the ospf domain (as E2 routes the default).
Routers B and C have uplinks to the ASBR in area 0
Router D has two uplinks to routers B and C but in ospf area 100.
Routers B and C are ABR's.

On router D with out setting the maximum-paths to 1 will equally load balance over the 2 paths (observed with sh ip route), crossing both routers B and C to reach the E2 routes behind the ASBR.

If I hard code maximum-paths to 1 (one route will remain in the ip routing table observed with sh ip route), how does the preferred path get chosen with details below in mind.

- The E2 metric is 20 through out the ospf domain
- The cost to each ABR is 64 from router D
- The cost from each ABR to the ASBR is 64

With these details in mind, everything ties, the cost in each direction for the E2 routes from router D is 148 (obviously I’m missing something). How is the preferred E2 path chosen?

Cheers,
Garry-

grichardson661 · ‎10-26-2010

Ok markom on the networking-forum.com has confirmed in my case the older router will win the election.

Peter Paluch · ‎10-26-2010

Garry,

I believe that the maximum-paths limit is not the choice of OSPF to decide which route out of several available paths will be installed in the routing table. The OSPF, as far as it goes, always identifies the entire set of best paths. However, the maximum-paths serves as a throttling mechanism to allow only a subset of these paths to be subsequently installed in the routing table. I would say that they are installed in the order they were learned or located by the SPF algorithm. In any case, the maximum-paths in my opinion does not influence the SPF algorithm - rather it affects only the resulting process of installing the already identified routes into the routing table. Therefore, I think that from the OSPF standpoint, there is no tiebreaker - because there is no tie for the OSPF at all.

Best regards,

Peter

grichardson661 · ‎10-27-2010

Hi Peter,

Thanks for your detailed response.

I can probably lab your theory regarding which route SPF discovers first. I'll look at this today.

Cheers,

Garry