Hi,
Is there any calculator to give an estimate of how many VMs (2 vCPU/4G memory) can be run in their full capacity simultaneously on a ESXi 5.5 Host (without any fail-over consideration that is no DRS and no HA) with the following detail:
Dell PowerEdge M630
36 CPUs x 2.294 GHz
2 Processor Sockets
18 Cores per Sokect
72 Logical Processors
524GB RAM
This is for visualizing Kiosks estimation.
Thank you
What is the definition of "full capacity"? Based on a pure allocation model with no overcommit, you can get about 18 VMs out of that system (CPU most constrained) which doesn't account for any ESXi overhead.
What is the definition of "full capacity"? Based on a pure allocation model with no overcommit, you can get about 18 VMs out of that system (CPU most constrained) which doesn't account for any ESXi overhead.
Thank you. This means that basically one VM per core.
By "full capacity" i mean maximum usage of CPU and RAM for each VM (each VM is a public kiosk).
It depend the CPU usage of your VMs and the performance do you want to have (Production or Test) who choose the Overcommit.
The monitoring and the design are very important !
Generally overcommit ratio of 3 for production, 2 for agressive VM and 1 for very sensitive production, after that you just check the CPU usage of ESXi generally must be lower than 50% and after the CPU ready of the most used VM of each hosts.
Very interesting, thanks. Is there any VMware document explaining these in particular?
Great article here from Dell https://communities.vmware.com/servlet/JiveServlet/previewBody/21181-102-1-28328/vsphere-oversubscri...
3:1 vCPU to pCPU ratio could be possible depending upon workload in your environment. So i read it from your servers that if hyperthreading is enabled you could have 72 pCPU so up to 200 vCPU could work....