Hi guys,
Recently, I been studied about metrics of vcenter.
So, my study was about development spreadsheet,
the spreadsheet is about, what happen when new vms will be adding into the vsphere cluster.
I considered assign the metrics per individual host ESXi, cpu usage, memory consumed, overhead memory in accordance vmware table:
also I counted the number of vcpus configured for all the vms and I counted the number of vram configured for all vms (this metric I used to understand to the worst case when the vms claim in 100% of vcpu and memory).
Ex. Host ESXi 20 cores (2 Physical Processor decacore) with 128 GB ram
In that hosts run 10 vms configured 4 vcpus and 16 GB vram
total 40 vcpus and 160 GB vram
Ratio 40/20 2Vcpus per 1 Pvcpu and 1,2GB vRam per 1GB Ram
I was read the documentation about cpu sizing wrote by scoot lowe, he mentioned a good ration when run into dell hardware 3 vcpus to 1 pcpu. Another documentation by vmware mentioned 10 vcpus to 1 pcpu. but this is not a context to discuss in this moment.
In configuration of vCenter metric is realy trust, in example when I create a custom cpu usage by month the peak high, low and average is reflected to reality.
So, I would like to understand, What are the way to mensure a impact to new vms or put more memory or cpu in VM exist.
What are the way I must analyze: the sum of resource of cluster or the sum of resource per individual host into the cluster