8600 ERS: Scalability Guidelines (OSPF Areas) 2

[rosenheim]

The 8600 scalability guidelines dictate a maximum of 5 OSPF areas per 8600.

1. Should area 0 be counted, or do they mean 5 excluding area 0?

2. Any idea what might happen if the limit is exceeded? We had a 8600 the number of which grew un-noticed to 9 areas (counting area 0 as well). Ours is a large LAN, and the 8600's tend to experience CPU utilization spikes. Once every few weeks a 8600 might reach 100% CPU state and hang forever in that state (until power recycled).

The areas issue had been corrected since, but no Nortel expert would relate it to the hang conditions (only a "maybe").

Could anyone shed some light on the 2 points mentioned above?
Thanks

 

[anthonyanderberg]

Page 45 in the Converged Campus Solutions Guide says the same thing, since they don't mention area 0 I'd assume it is included in the 5.

They also note that the numbers are version dependent so I assume they are concerned about CPU use during convergence. Did your high CPU levels happen during network convergence? Others on this group have reported high CPU use related to loops and other issues, it may be helpful for you to search for and look at those discussion.

 

[Alainv]

The numbers given by Nortel are save practise guidelines.
It won't say that the number has to be followed. Since those numbers take all other protocol and features in account.
If you need to be sure, you must take contact with the TRC but not the level 1 guys.

 

[shallowsc]

Hi, did you ever find the cause of the high cpu utilisation?

 

[rosenheim]

So far we have found one of the reasons. A server attached to a 8606GB card was generating large volumes of ICMP redirect packets, which on a non-E module are processed by the CPU (and not the ASIC).