Am I replying to Charles and Charles? To a virtual Charles and a real
Charles? Hmmm...not sure what layer this is.
I'm going to get into some depth on the reasons for configuration
guidelines, from the perspective of OSPF itself, not forwarding
performance. Just to keep this post of reasonable size, I'll start
with a discussion of single areas.
Apologies for any HTML. I don't know a good way to paste plain ASCII
into Eudora from a Word document, without doing lots of manual
editing. Slightly updated from Chapter 10 of my _Designing Routing
and Switching for Enterprise Networks_.
Area is an OSPF term for a part of a routing domain in which local
routing information is contained. In other words, change propagation
inside the area is at least partially hidden from the rest of the
routing domain. Only selected changes leak out, and there may be
restrictions on the external routes allowed into the area.
An area contains a set of cooperating routers that share a
synchronized and distributed topological database. Routers connected
to multiple areas have multiple databases. Even though this example
has a strongly hierarchical topology and addressing plan, the number
of routes is sufficiently small that summarization is not required.
An area is a basic element of calculating the load on a router. For a
link state protocol such as OSPF, in Area K, the CPU load for a
single routing table computation is based on:
1. The product of the number of intra-area routes in Area K
and the logarithm of the number of routers in Area K
2. The number of summarized or explicit inter-area routes
seen in Area K
3. The number of routes external to the routing domain seen in Area K
The first term is the actual Dijkstra computation.
Summarization affects the second term.
Stubbiness and external aggregation affects the third term.
But this deals with a single computation. The real CPU
requirements are going to come from how often it will be recomputed.
The most basic need for recomputation comes from unstable
routes. The guideline of 50 routers per area tends to assume a
2500-class router (i.e., a 68030 CPU) and a mix of decent LAN and WAN
links.
Do you have enough CPU? You want to keep the 5-minute
utilization under 50-60%, especially on routers that do not have
separate forwarding processors. If you don't have enough CPU,
decrease the number of links in the area or get routers with faster
CPUs.
>Hi, Charles,
>
>As usual, you reek of brilliance, and as usual, you have to respond to your
>own post since no one else has.
>
>This is an overly simplistic guideline, as it does not include factors such
>as memory: a 7xxx router with 4MB memory won't be as effective as a 3xxx
>with 128MB. Also, running other routing protocols and redistribution,
>distribution lists, access lists, static routes, etc. may also influence
>this up/down the scale.
>
>Otherwise, a good start... why haven't you gotten your CCIE yet if you are
>so smart?
>
>
>""Cthulu, CCIE Candidate"" <[EMAIL PROTECTED]> wrote in message
>8t9ukm$ba0$[EMAIL PROTECTED]">news:8t9ukm$ba0$[EMAIL PROTECTED]...
>> Hi, all,
>>
>> What do y'all think about the following as a general rule of thumb for
>> deploying OSPF? I pulled this information together from a variety of
>> sources, including Cisco TAC, this group, and the vending machine man who
>> gave me 50 cents back. I think following these guidelines would result in
>> designs that reduce the load on the routers (less databases and networks
>to
>> process) plus make for easier troubleshooting. Of course, I also think
>> the Loch Ness monster is real ( I saw her playing water polo with my
>> mother-in-law;)
>>
>> Anyways, your thoughts would be appreciated.
>>
>> Guidelines for Deploying OSPF
>>
>> Number of Areas Per Router
>> Router Series Number of Areas
>> 7xxx <5
>> 3xxx-4xxx <4
>> 2xxx <3
>>
>>
>> Number of Routers Per Area
> > <51 Routers Per Area
>>
>> Number of Networks Per Router
>> Router Series Number of Networks
>> 7xxx <150
>> 3xxx-4xxx <100
> > 2xxx <50
> >
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
