>hhmmmm.... I wonder which resource the authors of the Cisco Press book were
>using when the wrote that.
I seem to remember these from an IETF Standardization Report, IIRC
just before OSPF went to Full Standard the first time.
Unfortunately, my recollection this wasn't an RFC, but is in the WG
archives.
>
>When reading your post, I also pondered some of the contributing factors
>that I thought might determine answers such as you quote. I then went to the
>CCO design guide
>
>http://www.cisco.com/univercd/cc/td/doc/cisintwk/idg4/nd2003.htm#16749
>watch the wrap
>
>and read what this document had to say. the DG and I agree on several of the
>factors - CPU, router memory, number and types of interfaces, number of
>routes, stability of the underlying transport. obviously YMMV, as the
>numbers below indicate.
>
>BTW, the aforementioned link also states that an OSPF area should have no
>more than 50 routers and no OSPF router should have more than 60 neighbors.
Now, that latter is an interesting point. I happened to have to deal
with some internal sales hype in a competitive situation not long
ago, where the account team was screaming we should have more
neighbors than the competition. Now, it's unclear if this is
neighbors on all interfaces, or neighbors per medium. I happened to
work out the layout of a hello packet for a link with a 1500 byte
MTU, and, IIRC, there is only room for 47 neighbors in the packet.
The technical term, therefore, for someone claiming 60 neighbors on
an Ethernet is "liar."
Remember that lots of these guidelines are set up to work in
worst-case conditions. Would you assume the same limits would apply
to an all-2500 series area on unreliable third world links, or an
all-7200 (with NPE-300) on an optical network?
Now, one of the reasons people probe for these limits is that they
often become something that's used in a procurement decision --
bigger/faster being assumed better. But to come back to the problem
one is trying to solve, my basic assumption on drawing the boundaries
of an area is that, first, the routers in that area should form a
community of interest -- the applications are such that the traffic
would tend to stay in the area anyway. Drawing that sort of area
boundary may very well hold the numbers down.
Also, you have to be careful about what "router in an area" really
means. The largest "OSPF network" I worked on had about 2500
routers, all with OSPF enabled except for one division that ran IGRP.
Some out-of-the-box thinking simplified the situation enormously.
Turned out that out of the 2500 routers, only 400 really needed to
speak OSPF. The rest were all edge routers that could use
static/default routes to reach an OSPF router that actually had a
choice of routes. In general, the edge routers either had a single
path or a floating static backup.
>Nor should a router be in more than three areas, according to this document.
>I presume these rules of thumb were developed by Cisco engineers while
>working with major client installations. I know that Cisco actually does
>performance testing, and quite often the numbers we see in certification
>materials originated during the course of Cisco internal testing as well as
>field experience with clients.
>
>Chuck
>
>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
>Gareth Hinton
>Sent: Sunday, July 08, 2001 1:56 PM
>To: [EMAIL PROTECTED]
>Subject: Re: CID-How many routers in an area [7:11240]
>
>
>Excerpt from the Cisco Press BSCN book Page 177:
>
>OSPF Design Guidelines - Studies and real world implementations have led to
>the following OSPF design guidelines, as documented in OSPF Network Design
>Solutions:
>
>(Figures show Minimum, Mean, Maximum)
>
>Routers in a Domain - 20, 510, 1000
>Routers per single area - 20, 160, 350
>Areas per domain - 1, 23, 60
>
>
>Just to throw the cat among the pigeons.
>
>Regards,
>
>
>Gaz
>
>
>""Stephen Skinner"" wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> > guys,
>>
>> i am about to sit the CID (AGAIN) and have a question
>>
>> how many routers are allowed in an ospf area (max)
>>
>> i have seen 100 on the boson tests ....42 in some cisco docs and 50 on
the
>> CCo "designing ospf networks2....
>>
>> any idea which one the test wants....
>>
>>
>> cheers
>>
> > steve
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=11380&t=11240
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
