Skip to main content

Facebook's Open Compute Project shares plans for energy-efficient data center

We know, you've seen an awful lot of Zuckerberg and crew here lately, and the Facebook news just keeps rolling in. This time the social networking giant is doing some sharing of a different sort by offering public access to the specifications and best practices behind its new, more efficient data center in Prineville, Oregon. According to the company, the center, built in collaboration with AMD, Dell, HP, and Intel, has boosted energy efficiency by 38 percent while lowering cost by 24 percent. The information now available through the Open Compute Project includes technical specs and mechanical CAD files for everything from servers to building design. Basically, if you want to erect your own multi-million dollar Facebook-style data center, you've got the go ahead. Full PR after the break.


Facebook Launches Open Compute Project to Share Custom-Engineered, Highly Efficient Server and Data Center Technology With the World

PALO ALTO, Calif., April 7, 2011 /PRNewswire/ -- Seeking to transform the energy efficiency of global data centers, Facebook today launched the Open Compute Project, an initiative to share the custom-engineered technology in its first dedicated data center in Prineville, Oregon. This advanced technology delivered a 38 percent increase in energy efficiency at 24 percent lower cost for Facebook, and the specifications and best practices behind those gains will now be available to companies across the industry.

"Facebook and our development partners have invested tens of millions of dollars over the past two years to build upon industry specifications to create the most efficient computing infrastructure possible," said Jonathan Heiliger, vice president of technical operations at Facebook. "These advancements are good for Facebook, but we think they could benefit all companies. Today we're launching the Open Compute Project, a user-led forum, to share our designs and collaborate with anyone interested in highly efficient server and data center designs. We think it's time to demystify the biggest capital expense of an online business -- the infrastructure."

Inspired by the success of open source software, Facebook is publishing technical specifications and mechanical CAD files for the Prineville data center's servers, power supplies, server racks, battery backup systems and building design. This technology enabled the data center to achieve an initial power usage effectiveness (PUE) ratio of 1.07, compared with 1.5 for our existing facilities, which fall into the "best practice" category as defined by the U.S. Environmental Protection Agency*. Established by the Green Grid in 2007, PUE is an indicator of data center energy efficiency, and the lower the number, the better.

Facebook is releasing these designs as open hardware, aiming to encourage industry-wide collaboration around best practices for data center and server technology.

Advanced Micro Devices, Dell, HP and Intel are among the companies that co-developed technology with Facebook. In addition, Dell's Data Center Solutions business will design and build servers based on the Open Compute Project specification. Synnex Corporation will also serve as a vendor for Open Compute Project servers, offering fully integrated and tested solutions based on customers' specifications.

For Facebook's data center in Prineville, Open Compute Project hardware delivered:

Energy savings-The data center uses 38 percent less energy to do the same work as Facebook's exiting facilities. If a quarter of the data center capacity in the U.S. were built on Open Compute Project specifications, it would save enough energy to power more than 160,000 homes.

Cost savings-In addition to the energy savings, Open Compute Project hardware means data center infrastructure costs 24 percent less to build out than Facebook's existing data centers.

Materials savings-Servers use a vanity-free design with no paint, logos, stickers, or front panel – and are free of all non-essential parts. This saves more than 6 pounds of materials per server. In a typical data center**, this would save more than 120 tons of material from being manufactured, transported, and, ultimately, discarded.


Facebook is publishing specifications and mechanical designs for Open Compute Project hardware, including motherboards, power supply, server chassis, and server and battery cabinets. In addition, Facebook is making available its data center electrical and mechanical construction specifications.

Comments

Popular posts from this blog

Manual for Alienware M11x with Sandy Bridge confirms NVIDIA GT540M graphics

If the previous Alienware M11x R3 spec leak got you all giddy, then we have some good news for you: according to a manual dug up by one of our eagle-eyed readers, it appears that this year's M11x refresh will indeed be coming with second-gen Core i5 ULV and Core i7 ULV options, along with a faster DDR3 bus (1333MHz instead of 800MHz), a higher-res webcam (2MP instead of 1.3MP), an HD TrueLife LCD, and optional 3G / 4G mobile broadband. But of course, the real meat on this laptop is its graphics card, which turns out to be an NVIDIA GeForce GT540M with either 1Gb or 2GB of dedicated memory -- not bad for a laptop of this size. Unsurprisingly, no dates or prices are mentioned here, but given the early start of inventory clearance, it shouldn't be long before Round Rock reveals all.  Dell (ZIP)

IBM shows off Smarter Traveler traffic prediction tool

Traffic alerts on GPS devices may be old hat at this point, but there's obviously still plenty of room for improvement, and IBM now says it's managed to do just that with its new "Smarter Traveler" traffic prediction tool. Developed with the help of UC Berkeley's transportation group and the California Department of Transportation, the tool relies on predictive analytics software, GPS monitoring and sensors already on the roads to not only offer alerts, but build a model of each person's usual commuter route. Once the system is trained a bit, commuters are able to check out what's effectively a forecast of their entire route before they even leave the house, rather than simply be alerted to traffic problems before it's too late to avoid them. Head on past for the complete press release, and a quick video explanation of how it works. IBM, Caltrans and UC Berkeley Aim to Help Commuters Avoid Congested Roadways Before their Trip Begins First-of-a-K...

Sony Ericsson Xperia Neo delayed to Q3, Arc and Play facing limited supply due to Japanese quake

There hasn't been much good news coming out of Japan lately and this sadly keeps up with the unhappy trend. Sony Ericsson has officially bumped the broad launch (it's already available in limited quantities) of its Xperia Neo handset to at least July, explaining the delay as the result of "supply chain disruptions." Additionally, the Xperia Arc and Play devices, two other members of the company's new Android Gingerbread family, will be available in smaller volume than expected, at least for the near term. We guess that might go some way to explaining why the Xperia Play failed to reach some UK carriers in time for its April 1st launch date. Skip past the break for a statement from Sony Ericsson, who promises to be more explicit about the situation when it delivers its latest quarterly results on April 19th. As Sony Ericsson continues to assess the impact of the situation in Japan on its business, we have communicated to our operator customers and dis...