Sorry, questions not directly related to PVE. but I want to find out the answers.
1. As far as I know, it is possibly to assign 1 or more VCPU to VM using KVM.
But is it possible to control real CPU usage by every VM?
Can I config elsewhere that VM1 can consume no more than 10 % of real CPU time, and VM2 can consume no more than 20 % CPU time.
Also, what about privileges?
Can I config that VM1 has more privileges than VM2 and if real CPU is busy, VM1 will be first on queve?
The same question about openVZ, does version included in PVE 1.0 include CPU scheduler?
2. What about maxinum number of VM under KVM?
In order to get stable system, the max number of VM must be no more than the number of real CPU cores?
For example, If I have quad core xeon, it is better to start no more than 4 VMs on it?
1. As far as I know, it is possibly to assign 1 or more VCPU to VM using KVM.
But is it possible to control real CPU usage by every VM?
Can I config elsewhere that VM1 can consume no more than 10 % of real CPU time, and VM2 can consume no more than 20 % CPU time.
Also, what about privileges?
Can I config that VM1 has more privileges than VM2 and if real CPU is busy, VM1 will be first on queve?
The same question about openVZ, does version included in PVE 1.0 include CPU scheduler?
2. What about maxinum number of VM under KVM?
In order to get stable system, the max number of VM must be no more than the number of real CPU cores?
For example, If I have quad core xeon, it is better to start no more than 4 VMs on it?