From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Fri Feb 14 2003 - 18:26:28 GMT-3
At 12:56 PM -0600 2/14/03, Sam Munzani wrote:
>Team,
>
>I came across an interesting finding. I want to take everybody's
>opinion on this before putting anything in production. We are
>building an OSPF network that will eventually merge with company's
>main OSPF backbone network. The core group has assigned us ospf area
>number 555.
Have they assigned 666 to anyone? :-)
>
>When I configure all my routers with OSPF area 555(with no area 0 at
>all), it seems to be building up routing table. I always thought
>OSPF needs area 0 to function. Will this work of we add a non cisco
>device with area 555 configuration?
>
>What is the catch 22 in this configuration? I have started reading
>OSPF RFC to figure out all technical details.
>
>Thanks,
>Sam Munzani
The RFC doesn't forbid it.
As others have mentioned, NEVER number a single area OSPF
implementation 0.0.0.0, because that will require renumbering if you
ever have more areas. Keep area 0.0.0.0 simple.
One gotcha is to always write out all four octets of the area number.
Cisco will interpret area 1 as 0.0.0.1, but I've seen implementations
that will interpret that as 1.0.0.0.
Offhand, I've seen single nonzero areas work fine in Cisco,
Bay/Nortel RS, Timeplex, GateD, Zebra, Proteon and IBM
implementations. Although I don't know if Microsoft has rewritten
their code, their original OSPF was a port of Bay RS, as were some of
the early IBM routers.
IBM now has its own OSPF code. One thing to watch for, especially
with older IBM OSPF implementations, is the timers may be optimized
for SNA over OSPF rather than the standard OSPF defaults (e.g., 3
second rather than 10 second hellos).
This archive was generated by hypermail 2.1.4 : Sat Mar 01 2003 - 11:06:23 GMT-3