As discussed in previous posts like public vs private cloud, you are getting vCPU for your public cloud servers, which is shared among other VMs in the same hypervisor.
But how do you know if the hypervisor, the physical hardware CPU itself is being overpopulated?
What you need to do is:
- SSH to your server
- Run the top command: type top and press enter
- Take a look at the %st, yap, st stands for steal, the value should remain at 0%:
It’s impossible to monitor it by human eyes, for us we add it to Zabbix monitoring by using the Zabbix-agent with key:
system.cpu.util[all,steal,avg1] - Unit type: Float
Graph it out then you will see how much CPU time of the vCPU have been stolen.