RAM could be overcommitted (assigning VMs more memory than the host physical RAM) on an ESXi host, and this was one of the historical advantages of VMware products. However, just because you can, it does not necessarily mean that you should.
Now, in most cases, it is not recommended anymore because it will degrade the performance of your environment (for example, memory ballooning or swapping also have an impact on storage). So, the right memory sizing could be really important, not only for the workload but also for high-availability aspects (we will discuss this more in Chapter 13, Advanced Availability in vSphere 6.5).
Note also that there are more and more workloads that are becoming more memory bound—consider, for example, all in-memory databases. In those cases, you should design the hosts carefully and maybe consider a specific cluster for those workloads (depending on the numbers and the recommended practices from the software vendor).
For memory, the following table suggests some possible metrics:
|
Metric |
Description |
Why it's important |
|
Available free memory (average and least) |
Amount of physical memory available for allocation to a process or system. It is equal to the sum of memory assigned to the standby (cached), free, and zero page lists. |
This value indicates how much physical memory is not being used by your server. If you have excessive free memory then consider reducing the amount of RAM assigned to the server when moving it to a virtual host. Otherwise, a very low value defines a workload memory bound and, probably, you should increase the memory. |
|
Pages swapped/sec (average and peaks) |
Pages/sec is the rate at which pages are read from or written to disk to resolve hard page faults. This counter is a primary indicator of the kinds of faults that cause system-wide delays. |
This value counts the number of times per second that the computer must access virtual memory rather than physical memory. This number normally increases as available memory decreases. Too many pages/sec can cause excessive disk activity and create a disk bottlenecks and also indicates that a system does not have enough physical memory. |