Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member


I'm reading up on OSPF.

I understand the difference between STUB and Totally STUB areas, but... why use STUB?

Both systems generate a default gateway on the ABR.

STUB propagates LSA 3 + locally generated default route.

Totally STUB propagates only the locally generated defualt route.

Now, because both have the default route, why would I want to use STUB to provide LSA 3?

LSA 3 and default gateway seems to me a reduntant information.

Thanks for any clarification.



Re: OSPF: STUB & Totally STAB

More specific routes takes precendence over the default route. STUB feature would help in shorest path selection if you have more than one ABR and there may be some IA routes that have the optimal path via the 2nd ABR which, isn't the next hop for the default route.

Hence, STUB or totally stub selection depends on your topology/setup. Another thing to remember totally stub is only supported on Cisco equipment.



*Please rate all helpful posts.

New Member

Re: OSPF: STUB & Totally STAB

Thanks for your quick answer.

I.m reading that use more than one ABR on a STUB area is not a good network design as there are more default router injected, also Internal routers may take sub-optimal path to destinations and/or decide to do unwanted load balancing.

Any way, coming back to my original question, it make sense have this 2 type of areas when I have more than one ABR, but in STUB areas with a single ABR there's no point in using STUB against Totally STUB.


Re: OSPF: STUB & Totally STAB

Hi Friend,

Yes you are right in your concept that if you have "total stub area" there is no point to have "stub area" as it will definetely reduce the number of routes which router will hold.

But "total stub area" is a cisco prop. technology and when cisco implemented ospf in their ios code for their products it addded a new feature as they will not delete stub area feature because cisco products can be used with other vendors so keeping that in mind "stub area" was left but new feature of "total stub area" is added.

So if you have all cisco babies in an area with single ABR you can very well implement "total stub area".

HTH, if yes please rate the post.


Hall of Fame Super Silver

Re: OSPF: STUB & Totally STAB


I would like to respond to the comment about being not good network design to have more than one ABR in a stub area. This comment is fairly common and many people seem to believe it. But I would present a different way of looking at this issue. An area with a single ABR has a significant single point of failure. If the single ABR fails then all the routers (and all the users) within the area are cut off from the rest of the network. An area with at least two ABR has redundancy and if one of the ABR fails then the routers (and the users) still have access to everything in the network. It seems to me that occasional sub-optimal paths is a reasonable price to obtain the redundancy. I never think that it is the best design to on purpose produce significant single points of failure.

I would suggest an analogy as another way to examine this question. Assume that I live in a town where there is a single store that sells shirts for $15. Whenever I buy a shirt for $15 I would be happy because I am getting the best price for a shirt in this town. Or assume that there is another town where there are two stores. One of the stores is like the store in the first town and sells shirts for $15. But the other store sells shirts for $10. How many of us would say it was the optimum design to live in the first town and always pay $15 - we never pay any sub-optimal price.

In network design I think we should always try to balance the simplicity of a single alternative (we never take a sub-optimal path) against the benefit of having choices (we may sometimes take the sub-optimal path, but we have chances to take the optimal path and we have no single point of failure). I guess it is pretty clear what I think is the better network design.