[c-nsp] SNMP/Monitoring Cisco gear

[Gerald]

I'll ask this here, since I primarily have Ciscos deployed. I'm looking to
eventually automate and consolidate my SNMP monitoring of my cisco devices.

The things I want to address are:

- reduce redundant polling (automatically?)
- effectively and easily pool statistics (all T1s traffic are X; all
servers are Y;)
- automate changes (interface is down, graphing stops with no errors.
That's what network monitoring is for, not stats, new customer comes up
stats take the interface description and automatically update the stats
output. Company changes names, so modifying description on interface
updates everything else.)

I'm currently using Nagios for monitoring + mrtg for stats. With mrtg I
could script a new scan to force changes, but that still doesn't address
the redundant polling to get grouped data. Example: I have 2 primary
upstreams and in order to graph each one individually and then give a
3rd graph of the combined I have to hit the routers again with a second
query.

Is there a better way to monitor my cisco gear and provide the output like
mrtg AND make it flow with the changes or am I thinking too far ahead?

Programs I haven't looked at in detail yet that I'm vaguely aware of:
cricket, rrdtool, smokeping, anything that uses Cisco netflows.

In my mind this is something like rancid meets mrtg meets a brain that
knows how to use information that it already has instead of querying the
device again. I can script this into existence pretty shortly, but I
hate reinventing the wheel.

Gerald