What is the best practice in determining the actual power consumption of servers?

[Anonymous user]

I just received two Proliant DL380's each with two 750W PS's.  I am trying to, A. Figure out what their actual power consumption will be (not much more than idling I would say) and B.  What the best UPS would be.

My company has proposed either the smart ups 1500 or 3000, thought there are multiple options for either.  I like the idea of purchasing the 3000XL because we could potentially later on purchase an additional battery back up.

Any input on this topic would be appreciated.  I did find a document from HP stating that a standard DL380 will run at around 245W per PS.   This still seems a bit excessive as the backup estimate from this site states that the 3000 powering just these two servers would last maybe a minute.

[BillP]

It might not be a bad idea to see what our UPS selector says as well: UPS Selector Sizing Applications

[Anonymous user]

I use one of those "green products" they sell for measuring power consumption around the house for this purpose. Its a handheld device made by P3 International and is called a KILL A WATT...and only cost about $40.

P3 - Kill A Watt

Most server power supplies are autoranging and will run on 120 volt in addition to 208 volt so you can plug a pair of standard PC power cords into the server, have them go to a power bar and then plug the power bar into the KILL A WATT.

Here's a Youtube video:  How to measure power consumption (wattage/amperage) of any item or device - personal power meter $20...

Good luck!

[Anonymous user]

I used the UPS selector initially using total wattage as my criteria.  I think this could be more accurate if I knew what my actual consumption rate will be.  I then just selected my devices and had the calculator advise me of the best solution.  I will be purchasing the UPS based off of this calculation.  Though in truth I still feel as if I am making a bling purchase as I am not sure what my servers will actually draw.  I guess the lucky aspect of this is that my company is paying for this item and I am not.

@battman, that sounds like a helpful device. Thanks!

[BillP · Accepted solution]

Yes I agree. And Battman has a great suggestion. That tool is probably your best bet to measure the consumption at the heaviest use time of the servers.