[mrtg] Re: linux performance

[Volk,Gregory B]

> Does anyone have any suggestions on ways to speed up the 
> processing?  Is anyone out there monitoring an enterprise 
> this large?  

I'm using MRTG to monitor more than 75,000 targets these days. That
covers foundry closet switches, firewalls, routers, load balancers,
proxy appliances, UPSs, etc in three countries. If it responds to SNMP
queries, I'm polling it. As of this morning (after the daily config
rebuild which only takes about two hours) I'm hitting 9755 devices for a
grand total of 75656 targets. Yeah, that's a lot of polling especially
when you consider the usual two snmp queries per target.


I do it on fully outfitted Sun 480 (4 dual core procs with 16GB of ram).
The big thing that makes it possible is that I'm running on EMC dasd.
This just would not work without high performance storage. Even now I'm
beginning to hit the limit of my EMC. I think I'm sharing a disk or two
or three with another busy app and it's on my list of things to do to
talk with the dasd folks here about some optimizations.


To get MRTG to scale like this, I wrote a bunch of perl scripts to
spread the config files out across 563 MRTG daemons. That works out to
be 134 targets per daemon on average. That number is a little low given
the network latency that I usually deal with and I'm considering upping
it, but it's working okay right now. (The dasd contention is a bigger
issue for me than latency right now). 

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Archive     http://lists.ee.ethz.ch/mrtg
FAQ         http://faq.mrtg.org    Homepage     http://www.mrtg.org
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi