Score:1

Using (dual) ATS for dual PSU server?

us flag

There's two parts to my question:

  1. Does it make sense to hook up a dual PSU server to an ATS to speed up failover to the secondary circuit?
  2. If yes, does it make sense to use two separate ATSes to mitigate hardware failure of a single ATS?

I've experienced a situation at the application level, specifically the oVirt hypervisor on some server, which appears to cause the service to crash when power is momentarily lost from only 1 power supply.

This leads me to wonder if a server might not cut over from PSU A to PSU B fast enough in the event that circuit A fails, especially since I don't see manufacturers publishing any figures on their PSU switch time. If so, a mitigating solution might be to plug both PSUs into an ATS, which has stricter guarantees on the maximum switch time - hence question part 1. (This of course still doesn't safeguard against either PSU itself failing, but at least safeguards against circuit downtime during DC maintenance.)

I've only used ATSes for single PSU devices before, so it has never raised the possibility of having redundant ATSes. Which brings me to question part 2 - does it make sense to plug supply ATS 1 and ATS 2 both with the same pair of circuits A and B, and plug PSU 1 to ATS 1 and PSU 2 to ATS 2?

in flag
In many cases redundant PSU still requires multiple input to handle the load. And the load is shared, there is not switchover time here since they are always active. In regards to the crash I would investigate if the PSUs are going bad, old heated caps starting to leak. When working with redundancy, you should aim for redundancy all the way. 4 PSUs, 2 ATSs, 2 UPSs, but it's always down to how much is the redundancy worth in each step.
Zac67 avatar
ru flag
A *redundant* dual-PSU setup needs to supply the system with any combination of inputs. If that doesn't work, it's either not redundant (e.g. undersized) or broken.
mangohost

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.