Talk to me about UPS

Soldato
Joined
10 Oct 2005
Posts
8,706
Location
Nottingham
So, following a power cut last night I thinking that it might be a good idea to get a UPS (well maybe actually 2 in reality) so that things can be kept running during a short outage and at least be shutdown cleanly during a longer one. But this is something which I know very little about (most of my career I have dealt with enterprise servers in data centres where the UPS was something managed by facilities at a site level rather than being dealt with machine by machine).

The systems I have which I think would need to be covered would be
UPS1:
1x HP Gen8 Microserver with 2x2TB + 2x1TB running ESXi 6.0
2x QNAP 431X NAS with 4x4TB
1x QNAP 431 NAS with 4x2TB
1x QNAP 259 Pro NAS with 2x4TB
1x HP 1810g-24 24-port managed switch

UPS2:
1x HP Gen8 Microserver with 1x SSD + 1x1TB + 2x2TB running Ubuntu
1x BT Home Hub 6 (for WIFI access)
1x Netgear 8-port unmanaged switch

That would allow me to get onto the NASes and servers via WIFI from a tablet/phone/laptop and shut them down if necessary.

What sort of sizes and brands should I be looking at and what if I can get the actual power draw of the devices how do I do the calculation to work out how much capacity I need for how long? I don't you can get devices which you can put between the system and the wall to give an idea of the actual power draw ... anyone have any recommendations on those as well?
 
Last edited:
Soldato
Joined
6 Jun 2008
Posts
11,616
Location
Finland
None of them should have that high power draw, because HDDs don't need that much power and file servers don't do intensive processing and NASes have even less processing power.

Some UPS makers are very ambiguous giving only couple operating time numbers (full/half load) while others give proper charts showing battery life for various loads:
http://powerquality.eaton.com/ELP850DIN.aspx?cx=101

Depending on what kind OS/software those servers are running you might be able to get them to automatically power down when certain battery level is reached.
HID UPSes can tell PC their status/battery level in standardized way over USB.
 
Soldato
Joined
17 Dec 2004
Posts
8,696
Now I dont know even if i would take on my own advice as I love trying to find good deals, but hear goes...

My first ups was belkin, was about £150, 670watt and it kept my 150watt load going for about 20mins, I haven't had no problems, its never let my down when the mains power goes dodgy, plus its still going strong 8+yrs later. Anyway I have Just upgraded to another not well known UPS, Powerwalker, but I cannot really comment on this one yet, accept it was £400, 1800watt and it keeps my 150watt load going for 90mins and customer support is superb.

So its if you want to go with a well known/good brand one or not, but belkin ups has served me really well, people will say oh you want APC,Eaton, but just think what your actually paying for, is it the hardware or the actual name? I may of done a stupid thing spending £400 on not a well known UPS but I just couldnt justify spending double on a same spec ups but well known make... "Time willl tell, sooner or later time will tell" as the professor says out of Command & Conquer Red Alert 2.

I dont know if my 2 pence has helped at all really, but what your looking for is........ fast transfer time (4ms or lower) if possible, pure sine wave and get the highest wattage one you can afford, if your bothered about good runtime........ Ah yeah I said that I go for not well known ones but still dont skimp too much on the price even though... Research, Research, Research before you buy matey.

Ah yeah the pc spec at the bottom is the 150watt load.
 
Last edited:
Soldato
Joined
6 Jun 2008
Posts
11,616
Location
Finland
Grab one of those wall outlet adapters and get a measure on what you're drawing.
For non-linear/non-PFC equipped devices cheap power meters can have notable inaccuracies with actual power draw being 50% higher.

Text is in Finnish but result tables are rather obvious and first load is 75W incandescent light bulb, then capacitive load, small 15W desktop CCFL light and rest are with clear names:
http://jahonen.kapsi.fi/Electronics/PowerQuality/
Also Google Translate does actually pretty good translation.
 
Soldato
Joined
29 Jul 2011
Posts
15,603
Location
Near Northants / MK
For non-linear/non-PFC equipped devices cheap power meters can have notable inaccuracies with actual power draw being 50% higher.

Text is in Finnish but result tables are rather obvious and first load is 75W incandescent light bulb, then capacitive load, small 15W desktop CCFL light and rest are with clear names:
http://jahonen.kapsi.fi/Electronics/PowerQuality/
Also Google Translate does actually pretty good translation.

I guess we could sit here and guess what he's actually drawing then...
 
Soldato
Joined
6 Jun 2008
Posts
11,616
Location
Finland
I guess we could sit here and guess what he's actually drawing then...
Those NASses surely have some markings in power supply which would tell how much power they can draw when fully loaded with drives.
That would also tell about likelyhood of active-PFC, because regulations demand PFC from IIRC above 75W power draw devices.
Servers likely have modern PSU with active-PFC and even cheap meter would give reliable reading.
 
Soldato
OP
Joined
10 Oct 2005
Posts
8,706
Location
Nottingham
From memory I think the spec. page for the QNAP 431x is supposed to pull ~27W with 4x1TB drives ... but that's on paper so in reality could be different as they don't say if how heavily they were being hit when that measurement was taken (and nor what model disk so no way to compare with the 4TB drives I'm using). Quite a bit under 75W, although the power supply is supposed to be rated at 90W.
 
Back
Top Bottom