I'm having some confusion between the PRTG Hyper-V host sensor CPU totals and the standard WMI CPU Load sensor for the same server. There is a large discrepancy in their values.
For the exact same time period, the Hyper-V Host Sensor reports a maximum value of 31% for the Total CPU, whereas the CPU Load sensor peaks at only 2%. So, the two sensors are giving completely different values for the same CPU load.
Hyper-V Host: Server 2012 R2 Core running Hyper-V
PRTG Server: v. 14.1.8.1371 Installed on Server 2003
Could there be a problem with one of these sensors?
Article Comments
After doing some more research, it would seem there is a fundamental difference between these two counters in Hyper-V that I wasn't aware of....
The WMI CPU is reading from the performance counters of the Management OS, actually just another virtual OS ("Parent" partition or VM). So with the WMI sensor, you're not getting the host CPU usage, you're only seeing the CPU usage of the Management OS VM by itself.
The Hyper-V Host Sensor, however, is reading performance counters from the actual Hyper-V Hypervisor data, giving the real CPU usage, including that of all the guest VM's.
Hope this info saves someone else the confusion!
Mar, 2014 - Permalink
After doing some more research, it would seem there is a fundamental difference between these two counters in Hyper-V that I wasn't aware of....
The WMI CPU is reading from the performance counters of the Management OS, actually just another virtual OS ("Parent" partition or VM). So with the WMI sensor, you're not getting the host CPU usage, you're only seeing the CPU usage of the Management OS VM by itself.
The Hyper-V Host Sensor, however, is reading performance counters from the actual Hyper-V Hypervisor data, giving the real CPU usage, including that of all the guest VM's.
Hope this info saves someone else the confusion!
Mar, 2014 - Permalink