[cisco-voip] Convincing Management

[Mike Armstrong]

My system management philosophy has always been something like "the vendor 
goes to a lot of trouble to build these things, and I'm paying for all that 
effort, so I may as well use them", combined with "newer should be better", 
so I've generally upgraded whenever a new release or patch has come 
available.  I also feel that if a problem does arise, the vendor will 
probably be happier working on new stuff rather than old stuff.  I've been 
burned a few times in the last 40 years or so, but not enough to make me 
change my approach.  In the case of CallManager, we hadn't upgraded since 
3.3(3), and 4.1(2) had a couple of features that we were pretty excited 
about (desktop video, JAWS support), so the decision was fairly easy.  We're 
discovering more new features every day, some of which are useful or 
entertaining.  We have a tolerant, inquisitive user community.

I see 246 phones registered right now, and rarely see more than 6 or 8 calls 
in-flight.  H.323 gateway into a single PRI with 20 active channels.  As for 
hardware, we're using 1 1/4 GB in Compaq DL380G2s.  Publisher is an MP, 
Subscriber is a UP.  We put the 2nd processor in the Publisher because of 
long processor queues developing, in spite of low utilization.  Nobody's 
come up with a satisfactory explanation of that, but the 2nd processor fixed 
it.  RTMT, now that it's working, shows about 70% memory utilization on the 
Publisher, 55% on the Subscriber, with CPU utilization averaging around 10% 
on both.  Another GB of RAM is under $300 (2x512MB), but we see no need for 
it now, and it will only get cheaper.

Mike Armstrong
UF/IFAS CREC
Lake Alfred, FL