I am playing around nagios now (again) and it seems that I have found the nice way to monitor most of the stuff I need.
Monitoring Linux servers in not a big deal and a lot of custom plugins can be (and already) written for watching different stuff, but monitoring Windows servers were not so easy for me until this days. Now it seems that I found the way.
It is well known that SNMP under windows really sux (I mean the standard one), the thing that really caught my attention was Performance Counter. This can do the job well.
My idea was to integrate nagios, windows performance counters and rrd graphs. For this I have set up nagios, installed NRPE plugin on windows to be able to check it and added a plugin which is able to watch performance counters. Now I have everything: performance counters are monitored, nagios updates the state of services and does all that fancy stuff it is supposed to do and in addition, the output and performance data of the plugins is written by nagios to the named pipe from where my small perl script takes it and updates a bunch of rrd databases. A simple web GUI shows me all needed graphs.