At my company, I work closely with our Capacity Management team. Their role is to collect data points from various teams (storage, VMware, etc.) and determine what the current state of capacity is. That is way over simplifying what they do and the value they bring to the teams they work with is very high, and the recommendation to purchase or when to delay purchase can save the company a lot of money. I use vROPs to publish weekly reports with a number of metrics that the capacity management team uses to determine what clusters need more capacity, what clusters dont and if we can shift capacity from one that has extra to one that is desperately low. Even though the reports they get from vROPs provide many metrics, we key in on average remaining vms that can fit in a cluster. A few weeks ago, I upgrade vROPs from 6.1 to 6.2 and we found out that the calculations were just a bit off from what we have normally seen and that caused me to dig deeper into what actually is happening to determine how many remaining VMs can fit in a cluster.

vROps calculates the remaining vms based on the HA and capacity buffers set by the vROPs policy and vCenter Admission Control. The reason why we see a difference from 6.1 to 6.2 is there is not a change in the way vROps deals with capacity rather there is a difference in how vROps integrates with Admission Control in 6.2 and this would account for the discrepancy. Before we dig a little deeper into the buffers, lets take a look at Admission Control

vCenter Server uses admission control to ensure that sufficient resources are available in a cluster to provide failover protection and to ensure that virtual machine resource reservations are respected. There are three types of admission control. For this explanation, we’ll just focus on vSphere HA. vSphere HA ensures that sufficient resources in the cluster are reserved for virtual machine recovery in the event of host failure. Admission Control imposes constraints on resource usage and any action that would violate these constraints is not permitted like powering on a VM, migrating a VM to a host, or increasing the CPU and memory reservation of a VM. Only one type of Admission Control can be disabled – vSphere HA.

This is why when we look in vROps at remaining capacity in a cluster we see “HA (0%)” in the buffers column.

It’s important to note that the reason why we see the “HA (0%)” is because enabled (by default) in the vROps policy is to use High Availability Configurations check box; however for this example Admission Control is disabled in vCenter and because of that the percentage remains 0.

If Admission Control is enabled, one option is to define failover capacity by a percentage, this would be represented in vROps by a number in the “HA (0%)”

If you unchecked in the vROps policy the Use High Availability Configuration option. The “HA (0%)” is removed from the buffer column in the remaining capacity section.

The other part of the buffers is the “+10%”. This is an additional capacity buffer controlled by vROps policy and by default it was set at 10%. We do have the ability to adjust that buffer if needed.

If we do that math in any cluster we can see that the usable capacity that vROps is reporting for CPU and Memory (as well as disk) is correct. For example: usable memory capacity is physical host memory times the overcommit minus buffers. This is how we can confirm that vROps is accurate. I also believe that the remaining number of VM left to fit is accurate because it uses these numbers to determine how much remaining capacity is left. I’ve asked VMware support if they can provide me the math behind determining the number of average VMs left to fit in a cluster so I can double check the calculations. I’m waiting their response. The discrepancy in the reports I wrote about in the beginning of this post, is due to VMware making general improvements to their product and the integration with the Admission Control setting in 6.2. There is nothing specific in the release notes from VMware that was specific to this; however support told me that they have had other cases in relation to these settings in 6.2. Roomer has it VMware is working towards more interoperability between the two products in Vsphere 6.5 Im assuming vROps version 6.2 is prepping for that integration.