Can vmware share cpu between physical servers?
sattar
Member Posts: 48 ■■□□□□□□□□
[FONT="]Hi all,
I'm beginner in vmware world, I was reading an article and some part of it was vague for me. I just want to make sure I got the concept right.
For example let’s say we have three Physical servers (not vm, real machines) they are connected to a switch, and we install three ESXi on each physical machine, and windows server on top of three ESXi, and the servers that have a specific app are connected to the sql server to grab data or to import data in it. Is it possible to share their own cpu, ram, and hdd between them? (For example if the sql server needs more cpu and ram share the work load on other physical servers ram and cpu too?)
[/FONT]
I'm beginner in vmware world, I was reading an article and some part of it was vague for me. I just want to make sure I got the concept right.
For example let’s say we have three Physical servers (not vm, real machines) they are connected to a switch, and we install three ESXi on each physical machine, and windows server on top of three ESXi, and the servers that have a specific app are connected to the sql server to grab data or to import data in it. Is it possible to share their own cpu, ram, and hdd between them? (For example if the sql server needs more cpu and ram share the work load on other physical servers ram and cpu too?)
[/FONT]
Comments
-
cruwl Member Posts: 341 ■■□□□□□□□□No, the VM resources are limited to what is on the ESXi host. If each host has 4 CPUs at 2 GHz, the VM guest will only use the 4 CPUs on that ESXi host if they are assigned to it. If you assigned 6 CPUs to the VM guest it would not use 2 from another host. It would have to wait for CPU to be available out of the 4 on the ESXi Host and your VM will suffer for it.
This is the same for RAM and HDD unless using Shared storage. -
jibbajabba Member Posts: 4,317 ■■■■■■■■□□cruwl is correct. Whilst technically you could assign more than you have (which is overcommitment) - you'd have massive penalties as the CPUs would have to "schedule" the requests.
In a VMware environment adding more vCPUs than needed often results in poor performance.My own knowledge base made public: http://open902.com