Compute Resource selection when creating VM on ESXI

[zeebahi]

Hi everyone,

I am new to vmware so have some very basics question:

1)When creating a VM, one option is to choose a ESXI host that will provide compute resource for the VM such as processor.

Does this compute resource also include RAM ? if yes    is  it allocated out of ESXI HOST's total RAM?

2) Let say we have single socket dual core processor, assume hyper threading is not enabled.

Will  ESXI  see   1X2= 2 Logical processors or pcpu?

If hyper threading is enabled, then :

ESXI see 1 x2X2=4 Logical processors?

3)  VCPUS are mapped to logical processors in some ratio,  Ideally we want single VCPU mapped to single Logical CPU ( pCPU) to avoid contention. But we can also do over subscription for example:  4 VCPU are mapped to single Logical Processor ( pcpu), the draw back is 4 vCPU will be scheduled to single logical Processor , depending upon the load ,some VCPU will have to wait longer than others to be serviced by the logical processor.

Did i get it right?

4)  Let say we want to achieve following goal:

Bare minimum, a VM is given 2 VCPU, but in case of heavy load such as VCPU  hitting 190% utilization combined, i.e  each VCPU is at 95% utilization,   VM should be able to request 2 additional vCPU to meet the demand. Once the demand is gone, these 2 additional  VCPU should be returned to ESXI host.  Is it possible?

Appreciated!!

[NathanosBlightc · Accepted solution]

Does this compute resource also include RAM ?

Yes of course. it's include CPU, RAM, Disk, Network and any other required hardware for that virtual machine. In a moment ONLY one of the ESXi host in your virtualization infrastructure can provide all required hardware resources for your VM and if you migrate the VM to another host, destination host will handle this matter after migration completed.

About your 2nd question, ESXi is the responsible for providing vCPU for the VMs from its total physical resources (Sockets and Cores) but don' think about if you enable HT feature, it will increase count of available vCPU in the wizard of new virtual machine creation

Enabling HT does NOT mean you can increase total count of CPU cores. This feature just can help the host to better threads handling

Once the demand is gone, these 2 additional  VCPU should be returned to ESXI host.  Is it possible?

If you mean this hardware association happens dynamically with a intelligence mechanism, of course it's not possible!

You must calculate required resources for each VM, in a minimum usage, average demand and also high utilization. So if you faced to many times of high utilization CPU by a VM you need to assign more core to that VM.

---------------------------------------------------------------------------------------------------------

Was it helpful? Let us know by completing this short survey here.