On 2/07/2008, at 9:53 PM, Jasper Bryant-Greene wrote:
Monitoring isn't going to help much if they ignore/misinterpret the alarm, even after customers start complaining...
Well, it does if correctly configured. A well configured monitoring system has some intelligence of dependencies between network components, to limit the amount of noise generated when a big outage occurs so that situations exactly like this one do not happen. I build these sorts of things for my customers a bit, and while they can sometimes be complicated to set up initially, they get over it pretty quickly when at 4AM they get one or two messages about a radio or fibre or something being down and know they need to get in their car, instead of 50 messages about all manner of rubbish and have to have a cold shower to wake themselves up and then figure try out what's wrong once their head is clear. There is, of course, no substitute for well trained network monitoring/ management centre staff. From reading that bit of info squinting a bit with my head on it's side, this looks to me like a classic case of someone relying solely on alarms (ie SNMP traps) instead of periodic testing. That sort of stuff is old telco think. -- Nathan Ward