Can ESXi 8 do some kind of power saving?

[giu1io]

Hi,

I’m running ESXi 8 on one of those fanless no-name boxes with an Intel N5105. I haven’t yet had chance to check how much power it’s drawing from the wall, however it feels pretty warm even ant idle with no VMs running.

I was wondering if ESXi supports some kind of power saving on these kind of CPUs (maybe c-states or p-states?) or if it’s running all the cores at the maximum frequency all the time.

Thank you for your help.

 

[mrpasc]

Yes, ESXi can do some power savings if you allow to do so in BIOS. See https://docs.vmware.com/en/VMware-v...UID-4D1A6F4A-8C99-47C1-A8E6-EF3865603F5B.html
But have in mind: Type 1 Hypervisors are NOT designed to idle at low power consumption. They are optimised to deliver good performance at load.
If a server spent lot of time in iddle status then the server was improperly designed!

 

[giu1io]

Yes, ESXi can do some power savings if you allow to do so in BIOS. See https://docs.vmware.com/en/VMware-v...UID-4D1A6F4A-8C99-47C1-A8E6-EF3865603F5B.html

Thank you for your help! I’ve managed to check the power draw for the unit and it’s around what they were seeing in the STH videos (10-11W idle). I’ve tried playing with the setting you linked me to: setting “Low Power” mode shaves off 2-3 W of consumption but really hurts performance, so I’ll just leave on ”Balanced”. I tried booting Ubuntu from an external drive and that uses 9W at idle. I think the difference, all things considered, is fine.

But have in mind: Type 1 Hypervisors are NOT designed to idle at low power consumption. They are optimised to deliver good performance at load.
If a server spent lot of time in iddle status then the server was improperly designed!

Thanks for the heads up. I went with an hypervisor because I want to run multiple services on it, including something like Plex so I chose a CPU that was slightly more powerful. Of course I won’t be transcoding files all-the-time, but finger crossed I didn‘t design the server improperly since it’s my first one.