OSPF areas concept on a star topology!!!

CCIE_2011

Hi dudes .... how is every one. I have out of networking for more than 4 months due to some collage engagements.


Now i'm back to study BSCI materials....

I got an ospf Q, actually such question is related to a real world situation that I really didn;t understand.

Please checkout this topology.

All router are part of area 0. Other areas have two routers and a number of L2 switches.
routers A,B,C,D,E,F,G,H all have some subnets for users that are not part of area 0.


The Q:
will there be any benefit from having more than one area for such topology?
All the routers will have the whole network . They have the same topology.

I'd like to here from you, your opinion please


PS: sorry for the bad english

EdTheLad

This is the type of design you would choose for a huge network, if you were expecting to scale and have maybe 30 routers in each non-zero area.The idea to break up ospf into multiple areas is down to the time it takes ospf to perform a recalculation and the frequency of recalculations.
The ABRs can summarize routes but i really cant see you are gonna have that many routes to justify this design.You could start with a scaled down version and build towards this if required.Its like buying a quad core to play minesweeper.

CCIE_2011

Will the drawing is just to demonstration. We have in our network around 18 distributor L3 switches. It is quite big with many users and applications.
But i was thinking where the concept of areas will be beneficial to us.


Thanks for ur reply EdTheLad it was very informative.

I hope to hear more from others

networker050184

I agree with Ed that something this small would be fine in a single area. If you had plans for huge growth then I can see going with a design like this, but you probably wouldn't see much of a gain in performance.

redwarrior

We have something very similar to this, but we have moved routing down to the distribution layer and we have multiple campuses. We have noticed a performance increase that so far seems to be good enough to justify the complication. (Of course, we also have been simulatenously moving from routing only in the core to routing from distribution to the core, so that could also have explained the performance boost...apparently spanning-tree was insane before!)

tim100

CCIE_2011 wrote: »

Will the drawing is just to demonstration. We have in our network around 18 distributor L3 switches. It is quite big with many users and applications.
But i was thinking where the concept of areas will be beneficial to us.


Thanks for ur reply EdTheLad it was very informative.

I hope to hear more from others

It can be beneficial if there are a large number of routes. Areas 1,2,3 and 4 can be configured as stub no-summary areas to stop the propagation of summary LSAs into these areas because the ABRs will then inject one route of 0.0.0.0/0 into each stub instead of let's say 1,000 routes which is alot of LSAs. As EdtheLad stated, this type of design you would choose for a large network.