Green Data Centres
(or Centers for those who prefer North American spellings)
There's an interesting interview on O'Reilly and Associates' news pages [http://news.oreilly.com/2008/06/building-the-green-data-center.html] with Bill Coleman the founder of web apps company BEA Systems who recently set up Cassatt Corppration, about the challenges involved in making data centres more energy-efficient. He notes that energy efficiency is the starting point, and that application availability and resource repurposing matter a great deal to customers.
He also notes that 'data centers were never developed or did not evolve with any thought to power efficiency. You know power wasn't really a consideration as we started in the '60s, '70s, '80s, and '90s', and goes on to outline the different sources of inefficiency and waste.
It's well worth listening to or reading the whole interview, especially his comments on cooling systems and power cycling. One point Coleman makes is that a lot of the concern over turning computers on and off 'goes all the way back to when they used to have tubes in them' whereas today's reality is that turning systems off is a safe and sensible option. As he puts it, 'your mom taught you to turn off the light when you leave the room; well turn off the server when you're not using it.'
There's a discussion of the issue over on Slashdot [ http://news.slashdot.org/article.pl?sid=08/06/21/1538231 ] where the comments range, as usual, from the insightful (BigZaphod's note that 'software has an impact, too. Messy, heavy code takes longer to run, takes more CPUs, etc. Imagine how much energy could be saved if there wasn't so much code bloat!') to the jokey, as when Harry666 claims that the solution is '1. Get a data center 2. Paint it green'.
energy
| data center
| cooling
| data centre

1 Comment
One of the thik that we would like to solve in the next year student project is a metering device, specially suited to evaluate consumption of appliances in datacentres - an electricity meter with single voltage and a multitude of currents in a PDU housing. The measuring part should evolve from SMU device and we would like to be able to analyze relatively lowcost what is a real consumtion of each unit. If interrested, let me know. I will be also looking for any practical suggestions on post porocessing.