OSPF Areas Question

[vincehgov]

I have three areas - 0, 1, 2

I know that all areas much touch area 0.  But can area 1 touch area 2 so that it's basically a triangle?

see attached screenshot.

It seems to be working in my virtual environment.

[Joshua Schmookler]

I'm sure someone will come along and correct/clarify me on this, but:

I believe area 1 can certainly touch area 2, as long as they both touch area 0. The only time you could run into an issue is if the link from area 1 or area 2 were to go down.

You may want to configure virtual links in area 1 and 2 in case the link to area 0 fails.

[vincehgov]

The reason I'm asking is because I have a pretty small network.  However, I have three sites located in 3 geographically separate areas.  A is the "hq" site, and site B and C are satellite offices.  I have leased lines connecting all the sites together via a mesh topology.

I'm wondering if I should just configure it all to be area 0 or configure three areas with area 0 being at site A.  The problem would be if site A becomes unavailable, area 0 would be discontiguous. or non-existent.

[vincehgov]

Thanks for your help Josh.

[Peter Paluch · Accepted solution]

Vincent,

I do not see the clear area boundaries in your screenshot which is essential to have your query answered correctly. Let me therefore just provide a generic answer - I am sorry if it does not precisely answer your question.

If the link between Area 1 and Area 2 is not itself in Area 0 and if the Area 0 is not contiguous in the entire network, it (the link) won't be used by OSPF for routing. The routers may create an adjacency and exchange their topological databases but as both routers consider themselves to be ABRs (they have interfaces in at least two different areas), they will use only backbone summary-LSAs (LSA-3 and LSA-4) to calculate the inter-area routes, as mandated by RFC 2328, Section 16.2:

16.2.  Calculating the inter-area routes

        The inter-area routes are calculated by examining summary-LSAs.
        If the router has active attachments to multiple areas, only
        backbone summary-LSAs are examined.  Routers attached to a
        single area examine that area's summary-LSAs.

In other words, summary LSAs containing information about other areas exchanged via the link between Area 1 and Area 2 will be ignored during the routing table computation.

Simply said, the Area 0 may have an arbitrary topology but it must be contiguous, and all other areas must be attached to routers that have at least one interface in the (contiguous) Area 0. Otherwise, the network will be partitioned and most probably, no full reachability will be available.

Best regards,

Peter

[vincehgov]

Hi Peter,

I'm sorry, I forgot to touch up the screenshot with some labels.

All interfaces of R2 or R3 are either in Area 1 or Area 2. 

F1/0 and F1/1 of R1 is in Area 1 and Area 2.

All other interfaces of R1 is in Area 0.

R 4 is an internal router in Area 0.

So R1 is basically an ABR connection Area 0, Area 1, and Area 2.

I see what you're saying. Thanks Peter.

[manish arora]

Hi,

Just to input, I think in your topology you will be better off  making R2 & R3 as ABR's while R1 should be a Backbone internal  router Just like R4. Then you can  use the Link between R2 & R3  using a OSPF virtual Link and can solve the problems mentioned by Mr.  Peter of Backbone being partioned orlas type 3 or 4 being ignored.

Manish

[vincehgov]

Hi Manish, so you're saying that I should extend area 0 out to R2 and R3?  Thats a great idea.

The link on R2 going to R1 will be in Area 0.  The link on R3 going to R1 will be Area 0.  And the links connection R2 and R3 together will be area 0.  That would give R2 and R3 two interfaces in Area 0.

I've been using Eigrp all these years and I'm not use to having to design my network to work with my routing protocol.  It is interesting.

Thanks Manish! 

I really want to use OSPF instead of EIGRP so my network will be more "open" and so I can practice it.  But I'm afraid I might just be over-designing my network.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Vincent Gov wrote:
I have three areas - 0, 1, 2
I know that all areas much touch area 0.  But can area 1 touch area 2 so that it's basically a triangle?
see attached screenshot.
It seems to be working in my virtual environment.

Define working.

I would expect everything to be reachable, but would also expect traffic from the backside of an area 1 router would need to transit area 0 to reach the backside of area 2.  I.e. would not expect traffic to use the link between your area 1 and area 2 routers as transit.  Traffic to the transit network itself will likely find its way although be interesting to see what area 0 has for that subnet. I would suspect it would be listed as reachable via either area 1 or area 2 assuming cost is the same.

I think this agrees with Peter although I would be surprised if an adjacency formed between  area 1 - area 2 linked routers.  Did it or didn't it?

PS:

If you're trying to learn OSPF, try to keep in mind the purpose of areas is to allow scalability.  It's on ABRs where you can partition Dijkstra's algorithm and can control routes entering or leaving an area.  A very rough rule of thumb is no more than 50 routers to an area, but there's much to consider to "right size" an area.

I.e. for a small topology as you described you have, single area would likely be just fine.  Then your triangle works just fine.

NB: a single area doesn't have to be area 0.

[manish arora]

The general rule of thumb of having 50 routers per area is  a lay men solution , The no of routers in an  area completely depends upon  the following ( or company managers ;-) ) :-

1> No. of Links in an Area.

2> The memory & cpu of your Routers to handle the Link state Database + flooding to maintain these databases.

3> The stability of your Topology.

4> Use of summarization.

5> No. of summary LSA's enterning the area.

As for Vincent, even if it's a test lab for self learing , The best solution would be :-

1> Let R1 be the ABR , have R2 and R3 be internal router for a single non backbone Area, This way the intra area traffic will stay on the Link between R2 and R3 , whereas traffic for backbone area will use respective SPF route. In this case any Link failure would not cause any service loss. Even if the link between R2 AND R3 goes down  , that single area will be partitioned , but each partition can still find the ABR in Router R1 and ospf backbone will consider both areas as separate areas , will let the intra area traffic flow through the backbone as inter-area traffic.

2> As i mentioned earlier, extend the backbone to Router R2 and R3.

Manish

[Joseph W. Doherty]

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

Manish, nice summarization of most of the issues that I alluded to determining number of routers within an area.  I would also add bandwidth between routers.

PS:

". . . it's difficult to specify a maximum number of routers per area."

From:

http://www.cisco.com/en/US/tech/tk365/technologies_white_paper09186a0080094e9e.shtml#t40