[c-nsp] OSPF area design question

[Marcel Lammerse]

I guess. OSPF uses a LSDB *per* area. So it probably introduces more 
overhead.

On Aug 30, 2004, at 10:41 PM, Dan Armstrong wrote:

> We too have a similar situation.  We opted to make a whackload of OSPF 
> areas.
> I am very curious if this design is going to eat up some resource
> unnecessarily.
>
> I can't quite figure out why in a "real"  NSSA scenario that other 
> routers in
> the same area need to know anything about other routers in the stub 
> area,
> since the only path anywhere else is up to the distribution layer 
> anyway,
> which is handled with the default route that gets advertised down...
>
> Dan.
>
>
>
> On Monday 30 August 2004 16:37, Marcel Lammerse wrote:
>> Ok, if you have that area 1 with 15 routers. Would it be a good idea 
>> to
>> keep them all in one area, or would it make sense to assign 15
>> different area numbers and make each of them a separate area (NSSA in
>> this case). Because, I figured, an update from one of the router will
>> be flooded throughout the entire area which is totally unnecessary.
>>
>> I like to know whether the extra configuration and administrative
>> overhead is worth saving on unnecessary update floods and cpu cycles
>> processing them.
>>
>> On Aug 30, 2004, at 9:47 PM, James Hampton wrote:
>>> The way I'm reading this is that you have three hub routers connected
>>> like  points on a triangle, with each point having 15 or so spokes? 
>>> If
>>> this is the case I would make the top router(or the one in the 
>>> middle)
>>> area 0 and the others 1 and 2 or what ever numbering scheme you come
>>> up with. Than address each area with contiguous blocks so that you 
>>> can
>>> summarize and keep the routing table as small as possible. The spokes
>>> could be "stubby" sense they have only one way out.
>>>
>>> James
>>>
>>>
>>> On Mon, 30 Aug 2004 17:54:35 +0200 (CEST), Marcel Lammerse
>>>
>>> <lammerse at xs4all.nl> wrote:
>>>> Hi,
>>>>
>>>> I have a hub-and-spoke network, for which I'd like to use OSPF as a
>>>> routing protocol. The spoke sites will advertise their networks to
>>>> the hub and receive a default route from the hub.
>>>>
>>>> A common piece of advice in OSPF design literature, is to use
>>>> different
>>>> area numbers to prevent unnecessary LSA updates from flooding to
>>>> routers
>>>> that don't need the updates and to avoid the cpu processing 
>>>> overhead.
>>>>
>>>> The total network has some 50 routers.  There are 3 inter-connected
>>>> hubs
>>>> and some 15 routers per hub. The way I see it, I can do two things:
>>>>
>>>> 1.      assign a lot of area numbers to prevent the LSAs from
>>>> propagating
>>>>        through to routers that don't need them. However, this leads
>>>> to a
>>>>        relatively complex configuration.
>>>>
>>>> 2.      accept the, potentially small, bandwidth waste and don't 
>>>> care
>>>>        about the cpu overhead (we're talking 2600XMs here).
>>>>
>>>> Option 1 just doesn't seem worth it. Could someone provide some
>>>> advice,
>>>> experience or tips?
>>>>
>>>> Thanks.
>>>>
>>>> -Marcel
>>>>
>>>> _______________________________________________
>>>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>>>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>>>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>
>> ---
>> "..the price to pay for teenage sex is pretty high--
>> unwanted pregnancy, disease, and ending up with one ear
>> bigger than the rest because it's always cocked toward
>> the door in case the parents come home early."
>> 										- Michael Moore
>>
>> _______________________________________________
>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
>
>
---
"..the price to pay for teenage sex is pretty high--
unwanted pregnancy, disease, and ending up with one ear
bigger than the rest because it's always cocked toward
the door in case the parents come home early."
										- Michael Moore