Now, we reach the final component of capacity management policies, High Availability (HA) and buffers. The concept of usable capacity is not new to vRealize Operations 6.0, but there are some great improvements, especially around virtual machines.

First and foremost, we need to discuss usable capacity vs. actual capacity. In vCenter Operations Manager 5.x, it was possible to choose whether your remaining capacity was based on usable capacity or actual capacity. Actual capacity is the raw capacity of an object based on its physical configuration. Usable capacity, on the other hand, is the physical capacity minus buffers to reflect a real-world environment.

Since vRealize Operations 6.0, you now no longer have the option to select actual capacity, only which buffers effect usable capacity. This is a far better outcome, because I cannot think of one use case where an administrator would want the capacity to be based on physical raw capacity. Even in an environment where HA was not required, without any CPU or memory buffer, you do not want your ESXi hosts filled up to 100% where swapping and unresponsiveness are occurring.

High Availability and buffers affect your capacity estimates by removing a percentage of total capacity and therefore basing the capacity and time remaining badges using the smaller, usable capacity value. This is a common practice that administrators would previously do, since running any hypervisor (or workload, for that matter) at 100% usually encounters issues. An example of this is when memory becomes critically low on an ESXi host; the memory states progress to the state where memory compression and hypervisor swapping is occurring.

The following image is based on the vRealize Operations Time Remaining example diagram that shows the buffer's effect on capacity: