People like to run their vSphere Cluster Overcommit ratios at different levels, based on workload types, capacity available, and more. Generally, I see Cluster CPU Overcommit ratios around 4:1, which means there are 4-times as many vCPU allocated to VMs in the Cluster as there are CPU Cores on the ESXi Hosts that comprise the Cluster.
This is a bit different for Cluster Memory Overcommit ratios, which tend to be closer to 2:1, depending on workload types, capacity, and more. If your Cluster has 6GB Memory (based on the sum of the ESXi Hosts in it), and you've allocated 12GB to VMs in that Cluster, you are running 2:1.
Aria Operations calculates the CPU and Memory (and Disk) Overcommit Ratios against Clusters out-of-the-box, they are just metrics.
But what if we want to know the CPU and Memory Overcommit ratios after taking vShere HA into consideration (meaning should the Cluster lose an ESXi Host)? We'll use Super Metrics!
We could calculate the CPU and Memory available to each Cluster should it have one less ESXi Host, but we don't need to, Operations already gives it to us, in the form of "Usable Capacity". It shows like this in the UI.
We'll use these metrics to calculate our Super Metrics for Cluster CPU Overcommit Ratio (after vSphere HA) and Memory Overcommit Ratio (after vSphere HA). They look like this.
If you'd like to use these Super Metrics they are available here. I've used these Super Metrics to create a Cluster Overcommit Ratios Dashboard (and Views) detailing my environment.
This allows you to see several things:
Distributions of your Cluster Overcommit Ratios at the top
Overcommit Ratio formulas, including the Super Metrics
List of all Clusters including all 4 Overcommit Ratios for each
List of ESXi Hosts in each Cluster
VMs with the most vCPU/Memory allocated to them
If you'd like this Dashboard and Views, they are available here, enjoy!
Comments