VMware Cloud Community
RaviRaju
Contributor
Contributor

How to determine number of vm that can be supported on Esxi

Hi,

I have the following configuration of Esxi host:

Manufacturer : Dell Inc

Model : PowerEdge R710

CPU Cores :  8 CPUs x 2.2526 GHz

Processor Type : Intel (R) Xeon(R) CPU E5630 @ 2.53 GHz

Processsor Sockets : 2

Cores per Socket : 4

Logical Processors : 16

Hyperthreading : Active

Number of NICs : 6

Memory of 16 GB

DataStore : 250 GB

I would like to create 2 types of VM (one with 1GB RAM and other with 512MB/256MB RAM)

Can u guide me how to decide max no of vm's that can be supported on such a config

I usally create around 60 VM powered on, and monitor the esxtop on the hypervisor

Ref to http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=200100...

Can we create and power as many VM , so the CPU/Memory Overcommit is less than 0.5 or should it be equal to 0,so as ensure that performance is not compromised

It would be great if u can point me to some resources online, which i can refer to

Thanks,

Ravi Raju

Reply
0 Kudos
5 Replies
mpr4ul
Contributor
Contributor

At an average

Memory of 16 GB

DataStore : 250 GB

I would like to create 2 types of VM (one with 1GB RAM and other with 512MB/256MB RAM)

As per your configuration, we can run if datastore is enough

if all 1gb vms => 12-13 vms

if all 512mb vms => 20-24 vms

you can create the vms as per the Resource status of host

This is my assumption according past expierence

Reply
0 Kudos
ramkrishna1
Enthusiast
Enthusiast

Hi  RaviRaju

Welcome to communities .

it depend upon application you are running on it  .

And also user dependency mean number of user accessing that application.

"concentrate the mind on the present moment"
Reply
0 Kudos
RaviRaju
Contributor
Contributor

Thanks for the info

So u say that the no of VM supported can be obtained by brute-force method of monitoring the resource status on host as and when we create VM one by one.

In that case till what point (resource limits) can we create VM's, should we see to it that CPU & MEM overhead should not cross 0/0.5 in esxtop statistics of host

Reply
0 Kudos
RaviRaju
Contributor
Contributor

Thanks for the info

Im running a QNX os on each of the vm

Can u throw some light on user dependency mean number

Reply
0 Kudos
jklick
Enthusiast
Enthusiast

Hi Ravi,

There are two different variables we need to look at: memory allocated and memory used. How much you can overallocate your host will be somewhat dependant on how accurately the virtual machines are sized. As a couple of examples:

  • In a situation where you consistently provision your VMs with more memory than they need, it's possible to allocate 200% of the physical memory, but never use more thatn 50% of it. This is an extreme, but possible.
  • On the other hand, if your VMs with around what they need what they need, you could find yourself with a 90% allocation of physical resources and 85% utilization of those resources.

It really comes down to what your VMs are actually using. To better understand how to measure this, I recently did a blog post that talks a little about the metrics involved:

http://www.vkernel.com/reader/items/vsphere-memory-usage-metrics-task-manager

Let us know if this helps.

@JonathanKlick | www.vkernel.com
Reply
0 Kudos