advertising router field of OSPF LSAs Type 3, 4, 5, and 7

[yuchenglai]

Hi all,

I was wondering if anyone can provide insight on this.

In an OSPF area were routers A, B, and C are on different ethernet broadcast segments laid out as A - B - C, what would be populated in the "advertising router field" for type 3, 4, and 5 LSAs with router A being the ABR for the area?  From router C's perspective would be in the advertising router field for those LSAs.  Would it be router A or router B?

Any insight would be helpful!

[Peter Paluch]

David,

The advertising router field in LSAs always contains the OSPF Router ID of the router that originated the LSA. If the router A was the ABR for the area and if it originated and injected some LSA3 and LSA4 into that area then the advertising router would be set to router A's RID.

The LSA5 are generated by ASBRs, not ABRs. The advertising router field would be set to the ASBRs router ID.

Please feel free to ask further!

Best regards,

Peter

[yuchenglai]

Peter,

That was helpful!  It's interesting to note that if the router-id was not an interface advertised into OSPF there may not be a way to execute one "show" command to figure out the cost to the advertising router.  Instead, one would have to "walk" it back tracing the LSAs upstream to the advertising router to piece together the cost.  Am I correct on this?

David

[Peter Paluch]

Hello David,

Yes, you are absolutely correct! In fact, the Dijkstra algorithm in OSPF always performs this piece-by-piece work to compute the total cost. Using the show ip route to find out the distance to the advertising router would not be entirely precise, by the way - the total displayed cost would be increased by the cost of the interface that was used to initialize the Router ID.

Best regards,

Peter

[yuchenglai]

Peter,

When you said that it would not be entirely precise, do you mean that the cost of the route displayed by "sh ip route" is less than or more than it should?  I can understand that their would be a discrepancy in the cost if it was a loopback interface...but would there still be this cost discrepancy if the router-id is based on a physical interface?

Thanks for your time on this,

David