I have a pool of Horizon 7.1 linked clone desktops, 4GB memory, 2GB reserved, windows 10 64-bit client OS that I am running some testing on.
I have about 150 desktop on each ESXi 6.5 host.
I have noticed that when I reboot the whole pool of desktops at the same time, the vCenter metric for Active memory exceeds the physical installed memory. For example during the boot storm Active memory hits about 572GB even though the host only has 512GB installed. After that, when I am running my testing, Active Memory and indeed Consumed memory all seem to report normal expected values.
What is it about how Active Memory is measured makes this possible ?
I am presenting the results of the testing I am doing so I want to give an accurate explanation of what is happening that can produce this seemingly contradictory situation.
OS can use memory more than physical memory using virtual memory. It's not a flat memory use.
You need to understand virtual memory concept in OS.
Further more, the hyper visor will use some share memory block technique to use more memory.
Thanks for your reply h3nkY
I think I am ok with overall principal of active / virtual memory in this case and what you say gives a general explanation of how active could exceed physical.
Im my case, I suppose, I looking for an explanation of what this is happening with vSphere 6.5 that allows the Active Memory metric to exceed the physical installed memory. I have carried out extensive testing on earlier versions of vSphere, and never has the Active memory value exceeded the physical (even with similar workloads and virtual machine numbers).
Have VMware changed how they measure Active Memory in this version I wonder, does it now more accurately or less accurately reflect Active Memory usage than on earlier vSpehere versions.