Server Fault is a question and answer site for system and network administrators. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I know that VPU stands for Virtual Processing Unit (versus CPU), but what does that mean exactly? Is it just the new terminology for referring to processor resources in a virtual server versus a dedicated (physical) server?

My understanding is that VMWare can apportion cores in a processor across virtual servers. Can VMWare also allocate virtual cores that exist due to hyperthreading?

Is a VPU a shared core (i.e. shared across other hosting accounts)? Or does each VPS typically get 100% of its allocated VPU?

Is a Cloud Hosting VPS different from a traditional VPS from and end user's perspective?

I'm still a little fuzzy on these concepts, and I'm trying to evaluate different hosting plans.

share|improve this question
up vote 4 down vote accepted
  1. It means that your VM is assigned a number of virtual CPUs, the OS in that VM 'sees' that/those virtual CPUs and can schedule work against it/them. Of course they're actually mapped to real CPUs but the virtualisation/time-slicing mechanism hands out real CPU processing capability to the VMs as required.

  2. You could say that yes, though I'd have said 'virtual server versus a physical server' myself.

  3. Yes but VMWare doesn't present sub-core units to VMs, if you have a 4 vCPU VM then its work will be carried out by one or more real physical CPUs/threads.

  4. Yes it's shared (unless specifically, and stupidly, specified to be dedicated).

  5. No not really, you just have no idea where your VM is in the world.

share|improve this answer
    
Thanks. So is a traditional VPS core shared as well (expanding on question #4)? Will a shared VPU suffer poorer performance versus a (stupidly) dedicated VPU? – JohnB Jun 15 '11 at 16:02
    
a) yes and b) it depends on load – Chopper3 Jun 15 '11 at 16:20
    
b) is pretty vague! – JohnB Jun 15 '11 at 19:11

Process control or monitoring sysems benefit from stay resident implementations. When a program/function is going to run 120 times per second, then the overhead of reloading program and data is not justified given the cost of cpus.Each application should not be approach with one hammer to solve all problems. Several of the applications we install gain no advantage of a multi-core cpu after four cores. Memory clashes are the problem. Solving large matrices of 15,000 equations with 10,000 changes is now limited to execution every fifteen minutes. The systems need solutions every ten seconds to prevent system collapse. I suggest that each application be examined instead of one solution for all.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.