OSPF stub areas practical uses

[imran.moulvi]

I was trying to find a document that explained how and where different types of OSPF stub areas can be used(like spoke locations, partners etc).

If anyone has such an explaination, please share.

[Thotsaphon Lueangwattanaphong]

Hi,

   It's some kind of filtering external LSAs into stub area to reduce area calculation when external routes flap. You are good to go : http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094e9e.shtml#t31

HTH,

Toshi

[imran.moulvi]

Hello Toshi,

Thank you for that link.

It would be helpful to get some recommendations of where/when each type of stub area should be used.

[Thotsaphon Lueangwattanaphong]

Hi,

    It depends on your design. You might use just one area for the whole network. You might inject external routes into HQ and design branch sites for stubby areas.

HTH,

Toshi

[Craig_Baum_2]

hi, I would consider such variables as whether the stub area has a contiguous block of addresses or are they varied. How many exits are available now and likely to be available in the future.

EG a small office location with say one WAN link with a /24 block, ideal stub with default route out. However if that site is likely to grow, maybe a few more /25 networks, a 2nd resilient route out then the use of a more dynamic update is required.

Do you have any servers in the area likely to cause application problems if an ip address within the stub is lost.

For me a remote branch office with low employee numbers, no servers etc and single link out is ideal for stub area.

Sent from Cisco Technical Support iPad App

[Thotsaphon Lueangwattanaphong]

Hi Craig,

       Good job.

5-point for you

Toshi

[Craig_Baum_2]

Thanks Toshi.

Should also have added how powerful your remote router is. A small branch router probably doing lots of local dns/dhcp services etc. can get quite overwhelmed by large OSPF database/updates, bandwidth on slow links etc..

I learnt the hard way. We started by throttling back the OSPF hellos and even recalc timers and then summarising but in the end switched to stub areas and the migration was a pain trying to do it in service because the OSPF neighbours were of course dropping due to non match of stub flag until all in the area were done!

Sent from Cisco Technical Support iPad App

[imran.moulvi]

Thank You Craig....

I also want to know about the NSSA and Totally NSSA types.

As I understand, NSSA areas dont have a default route, so the only reason is redistributed routes.

Can you advise where we would use these areas(NSSA and totally NSSA(Cisco proprietory).

Thanks.

[Craig_Baum_2]

So your NSSA is practical if your stub area has a boundary with another routing protocol such as RIP or IGRP.

These routes cant be summarised and a type 5 LSA generated in a stub area (which removes type 4 and 5) but by making it a NSSA the boundary router can inject type 7 LSAs..

Thus the OSPF cloud can learn of routing changes in the non OSPF area attached to a Stub area.

Totally stubby the type 3 LSAs are removed and replaced with a default route from the ABR.

Sent from Cisco Technical Support iPad App

[Rzodkiewka]

Hello. I am wondering what is the advantage of Stub Area of having additional routes (belonging to other areas) over Totally Stubby Area which has only default route and all routes inside its area.

Can you give me an example where Stub Area can be used while Totally Stubby Area cant ?

[Rolf Fischer]

Artur,

if the Stub Area is connected to the Backbone by only one Area Border Router, there is actually no advantage. It's like leaving a dead-end road: All you have to know to leave it is the direction.

However, when multiple ABRs exist, some Inter-Area destinations may have a shorter path through ABR-1, other destinations through ABR-2, and so on. This is the information you'll inevitably loose in a Totally Stubby Area: For all destinations outside the Area, the same (best) default-route will be used, which could result in suboptimal paths for some of the destinations.

HTH

Rolf