On 2/11/10 8:15 AM, Edward Ned Harvey wrote: >> 5 amps per node! Woweee! Those would be some beastly servers. I would >> expect a reasonable server as described by Ski to be closer to 2 amps. > > The best thing to do is to buy something like a Kill-A-Watt, and put your > server under load, and measure it.
I've done that around the office since I'm on the "green team" here. However, the data center is 240V instead of 120V. Is there an equivalent to the Kill-A-Watt for server-type connections which use C13-C14 power cords? By equivalent I mean something which doesn't require an electrical engineer or over $100 in purchase price. >8^) > Record the VA, the W, and the A. I > normally inflate all of these measurements 10-20% just for the heck of it. > Keep a spreadsheet of all the servers, make sure you don't exceed 80% of > your power circuit A's, or 90% of the UPS VA's or W's. > > I do this for every server I have, and the results vary enormously per > server. Late model Dell servers are extremely power efficient, and your 2A > estimate might be reasonable, even generous. However, I have some old model > Dells that require 460 VA (4A) and I have a recent Sun fileserver that > requires about 550VA (5A). Where you get charged for power, buying new machines regularly pays for itself pretty quickly, compared to places where you only get charged for space. We are currently charged per circuit, so we have to get smaller, higher efficiency machines to take advantage of consolidation. Otherwise we'd have to get the electric re-run with lower amperage circuits. Pleasantly, we're still buying new machines, because new with 5-year warranty is cheaper than annual maintenance on machines older than 5 years. _______________________________________________ Tech mailing list [email protected] http://lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
