PRTG can monitor Windows servers via WMI or SNMP.
We don't want to use SNMP on Windows Servers because: - It's deprecated - It's insecure
I'm having a lot difficulty getting WMI to behave in a consistent manner. VMs that are on the same subnet and are configured almost exactly the same, with the same admin groups and firewalls turned off are not being auto-discovered consistently. Some VMs show 8 discovered sensors and others show as 4 even though they all have the same hardware (cloned VMs). WMI also has a limit of how many WMI connections can be initiated a given time from a probe perspective.
In order to maybe solve the issue of security and WMI limitations, I feel like going the route of making all Windows Servers a remote probe. I think this is crazy, but I also think that this is the best way for WMI discovery/polling to be consistent and fast. Each remote probe is just in charge of polling it's own WMI. No worries about permissions and windows/external firewalls. Just open one port for remote-to-core traffic. This is kind of a hack-around for a agent-based monitoring solution.
Deployment can be scripted out and PRTG probe updates can be managed by Shavlik. So I'm not worried about that those things. I've just never seen this mentioned on kb.paessler and maybe there's something glaring that I could missing for this kind of deployment (max remote probe limits, management of multiple remote probes is a headache in PRTG, etc).
Any thoughts on this design?