i heard a plausable explanation as to what happened on sunday last. engineers took the ups off line to perform scheduled maintenance. the stand by generator took over as planned and work prceeded as normal then !!!!!!!! the generator shut down from a faulty oil pressure switch . alternate power was slow to be restored as no one knew how to manually change the sytem over . e&oe this a theory and in no way is an official explanation .
I heard that it was power related as well, all the fancy technology in the world doesn't help if you ain't got no juice baby!
Reminds me of a interesting story someone relayed to me about a data centre which had backup generators, and spare main diesel tanks positioned further away from the data centre itself so as to not blow the data centre up.
They tested their generators once a month like good little geeks, then the power went out for real, for a reasonable amount of time, and the smaller tank of diesel positioned next to the generator ran out of diesel, and the pump which pulled it from the main tanks was not working because it was plugged into the non-backup powered circuits, so the generator ran out of juice and stopped.
All worked fine during testing because they never actually disconnected from the main power grid, so the pump had stayed powered during the monthly drills.
I wonder what caused the IBM data centre power outage? If servers corrupted, then their SAN or possibly their vm layer went down hard... Should have hit the UPS's and done a safe shutdown.
I guess the people which know arn't talking, under strict instructions from lawyers. :-)
You know, google's idea (which they managed to keep secret for a amazingly long amount of time BTW) is to have two sealed lead acid 12 Volt batteries built into the actual server, so each server had exactly enough power to do it's shutdown without scaling or efficiently problems. Of course to achieve this, they had specially made 12V rail motherboards made.
I wish IBM (and HP) would built the option to slide small UPS's into servers, they could slide in like the hotswap redundant power suppliers.
Tyler - Parnell Geek - iPhone 3G - Lenovo X301 - Kaseya - Great Western Steak House, these are some of my favourite things.
I wish IBM (and HP) would built the option to slide small UPS's into servers, they could slide in like the hotswap redundant power suppliers.
That is very unlickly as running a UPS infront of another UPS ie infront of a the Data centres main in line UPS is not a good idea as it introduces all manner of nasty hormonics. If your DC is onto-it they won't allow you to.
Even just asking to put a little 1000VA UPS bought from DSE infront of a several $100,000 dollar enterprise UPS will also likely offend the Data Centre's infrustructure manager :)
exportgoldman:
I wonder what caused the IBM data centre power outage? If servers corrupted, then their SAN or possibly their vm layer went down hard... Should have hit the UPS's and done a safe shutdown.
Data Centre UPS's aren't like the little residential or SOHO ones, you don't get to plug a USB cable into them and have your servier notified when its running on battery.
Are you subscribed to our RSS feed? You can download the latest headlines and summaries from our stories directly
to your computer or smartphone by using a feed reader.
ID Verified
Trusted
