Hi,
I have been looking for a good explanation of all the metrics available in the monitoring charts of vsphere, but so far did not find it, so I am asking here.
My problem is the following.
I am asked to produce for our application (running in Redhat 7.X) a TPS/Ghz ratio, in other words, for 1 GHz given to a VM, how much TPS will I get.
So I looked in the monitoring charts, which gives 5 metrics in this area.
Here is what I see in the VM monitoring tab when it is under high load.
- CPU demand: 20300 MHz (almost constant)
- Entitlement: 17581 (totally stable)
- Demand to entitlement ratio: 115%
- Usage : 100% (totally stable)
- Usage in Mhz: 20400 MHz, (almost constant)
The guest OS gives me around 96% of CPU used (using top)
My VM is configured with 8 vCPUs, the host a 10 hyper-threaded physical cores running at 2397 MHz, so 20 logical cores.
I did not set any CPU or resource reservation for the VM, all VM settings are default.
1) Where the entitlement value of 17581 MHz comes from ?
2) Am I right to conclude that the real nb of MHz my VM is getting is 20400 MHz
3) If yes:
- Why is it slightly higher than the demand ?
- Am I right to say that each vCPU is in fact running at 20400/8 = 2550 MHz, so higher than the physical CPU frequency ?
4) Am I correct if I say that overall, the actual CPU used in this case is 96% of these 20400 MHz ?
Thanks in advance for any explanation.
