Pretty cool: Lower power usage models?

verulian

Renowned Member
Feb 18, 2019
203
38
68
45
  • Like
Reactions: mike the newb
Hello,

Proxmox is on Linux and opensource so you free to do what you want and make described in the articles possible.
 
  • Like
Reactions: Johannes S
Hello,

Proxmox is on Linux and opensource so you free to do what you want and make described in the articles possible.
Do you not think it would be worthwhile to share such information for the community so that I myself don't end up with a nice Frankenstein's monster-style creation that others cannot enjoy? Yes, the beauty of open source: share and others can enjoy in the fun if/when they have time.
 
  • Like
Reactions: mike the newb
I haven't read these yet, but I thought they sounded quite interesting if they could could perhaps translate to seeing Proxmox being able to somehow integrate a low-power usage model that might make some (seeming small) tradeoffs for deeper energy savings potentially:

What kind of power usage are you seeing currently? I'm running older SuperMicro servers, one with MS/hyper-v and 5 virtual systems, and the other running PM, albeit with minimal load since I'm just learning, but the first has a peak consumption of 80w, and an average of 50w. The second is even lower.
If you're running 24/7, I'm not sure how much lower you could possibly get...even though a newer system might theoretically be better than my 10+ year old systems.

I'd love to hear what comes out of this thread though.
 
  • Like
Reactions: Kingneutron
What kind of power usage are you seeing currently? I'm running older SuperMicro servers, one with MS/hyper-v and 5 virtual systems, and the other running PM, albeit with minimal load since I'm just learning, but the first has a peak consumption of 80w, and an average of 50w. The second is even lower.
If you're running 24/7, I'm not sure how much lower you could possibly get...even though a newer system might theoretically be better than my 10+ year old systems.

I'd love to hear what comes out of this thread though.
I’m running three Dell PowerEdge R710s (not clustered), 128GB RAM each, dual Xeons, with Proxmox. Across the trio I host ~10-20 LXC containers and 8-15 VMs. Per box I see ~350W peak, ~250W average, and ~160W idle (single 570W PSU, mostly SSDs). Lab‑wide that’s roughly ~1.05kW peak, ~750W average, and ~480W idle... Tweaks like L‑series CPUs, a single PSU, and BIOS power‑save MIGHT shave ~20–40W per node (not sure I'd want to do that though)... but my systems will never sip like your 50W SuperMicros!

So yeah... I definitely would like to explore these options at some point either myself or if someone else beats me to it.
 
Last edited:
I haven't read these yet, but I thought they sounded quite interesting if they could could perhaps translate to seeing Proxmox being able to somehow integrate a low-power usage model that might make some (seeming small) tradeoffs for deeper energy savings potentially:

There are very different use cases, for TELCO/Bussiness - for virtualization needs reliability and constant performance and redundancy. There are many topics in here, talking about "Power Saving" causes weird firmware/BIOS/kernel bugs/hangs/reboots/freezes.

If you want such features, then I think you don't need local infra, go to the cloud providers.
 
I’m running three Dell PowerEdge R710s (not clustered), 128GB RAM each, dual Xeons, with Proxmox. Across the trio I host ~10-20 LXC containers and 8-15 VMs. Per box I see ~350W peak, ~250W average, and ~160W idle (single 570W PSU, mostly SSDs). Lab‑wide that’s roughly ~1.05kW peak, ~750W average, and ~480W idle... Tweaks like L‑series CPUs, a single PSU, and BIOS power‑save MIGHT shave ~20–40W per node (not sure I'd want to do that though)... but my systems will never sip like your 50W SuperMicros!

So yeah... I definitely would like to explore these options at some point either myself or if someone else beats me to it.


Yep, that'll suck up some power. I'd be looking for power savings too :)
 
  • Like
Reactions: verulian
There are very different use cases, for TELCO/Bussiness - for virtualization needs reliability and constant performance and redundancy. There are many topics in here, talking about "Power Saving" causes weird firmware/BIOS/kernel bugs/hangs/reboots/freezes.

If you want such features, then I think you don't need local infra, go to the cloud providers.
Yep. Power savings is all fine and well, but for a SERVER you want reliable uptime. Then you gotta do your research into power supplies and system reviews, and hook it up to a Kill-A-Watt or similar meter. And if you don't need a huge RAID running 24/7: get a few large disks, copy daily-use data there, and only RAID on the weekend.

Powertop package can help, but in the BIOS you probably want maximum-run settings for stability. Homelabbers gotta remember, it's not a desktop.
 
  • Like
Reactions: Johannes S and UdoB
Let's actually find out vs having speculative pontifications... A lot of the naysaying here is not feet-on-pavement experience, but rather expressions of belief.