I am curious what the implications are of switching monitoring for around 500 Windows nodes from SNMP to WMI based nodes for monitoring in Orion? I know WMI is a significantly slower protocol so I am curious if this will have negative impacts on polling completion... or anything else I may not have thought of.
Thanks in advance for any feedback!
Last year I did a fairly in-dept analysis of the differences (in terms of load on systems) between ping-only, SNMP, and WMI. You can find it here:
Hope that helps!
As Leon Adato's very in-depth analysis confirms, there is very little difference between SNMP and WMI as it relates to the monitored servers. I would argue that the difference is so insignificant as to not be measurable. The Orion server however does undergo some additional load, though certainly not enough enough to discontinue use. This additional resource consumption can usually be offset by adding additional CPU cores and memory to the physical or virtual machine where Orion is running. The only major difference between WMI and SNMP is its performance across bandwidth contentious high latency links. WMI is primarily a LAN based protocol that does not handle +500MS response times very well. SNMP, being UDP based is very latency friendly.
Our biggest issue is there is no alert for when WMI is unable to Poll a node. So we may have a node that quits collecting data but is up and we do not know. With 5500 nodes, that is no fun.
Also, WMI uses a range of port rather than a port so, we had to work with the firewall and AD guys to limit the WMI range and then open it through the Wall of Fire.
The simplest solution would be to create an alert for WMI nodes that notifies you when a volume is in an unknown state for longer than "x" amount of time. Node status is determined by ICMP, no differently than SNMP managed nodes so I'm not sure how the problem you describe is unique for WMI managed nodes? These forums have plenty of threads where people are discussing similar issues when the SNMP daemon on their device stops responding. The solutions that address that issue are equally applicable in your scenario as well.
What you have mentioned is correct - to the best of my knowledge most suitable would be SNMP, but then if you have decided to go with WMI it shouldn't affect Solarwinds unless and until you have frequent health check up placed on your servers(Solarwinds Servers Health check + Solarwinds DB maintenance). WMI would definitely increase the load not just on Solarwinds but on the node side as well(the node that is managed by Solarwinds).
SolarWinds solutions are rooted in our deep connection to our user base in the THWACK® online community. More than 150,000 members are here to solve problems, share technology and best practices, and directly contribute to our product development process. Learn more today by joining now.