News

Facebook To Refresh Servers, Give Specs to Open Compute Project

• By John K. Waters
• 03/24/2017

Facebook will refresh its entire server hardware fleet, providing design specifications to the Open Compute Project (OCP) Foundation. The preliminary specs are available now on the OCP project site.

In an announcement made at the OCP Summit 2017, Facebook exec Arlene Gabriana Murillo said: "With people watching more than 100 million hours of video every day on Facebook, 95-plus million photos and videos being posted to Instagram every day and 400 million people now using voice and video chat on Messenger every month, we continue to innovate on our server hardware fleet to scale and improve the performance of our apps and services."

The Facebook server fleet refresh will include: the Bryce Canyon high-density storage server, the company's first major storage chassis upgrades since Knox (Open Vault) was released in 2013; Big Basin, Facebook's new GPU server, and the successor to the Big Sur GPU server; The Tioga Pass dual-socket motherboard, the successor to Leopard (the company's first dual-CPU server to use OpenBMC after it was introduced with its Mono Lake server last year); and Yosemite v2, a new version of Yosemite, the company's first-generation multi-node compute platform.

Facebook virtually launched the movement toward shared design and development of more efficient datacenter hardware when it established the OCP in 2011 with a group of partners that included Intel, Rackspace, Goldman Sacks and Sun Microsystem's cofounder Andy Bechtolsheim. The project grew out of an internal Facebook effort that began two years earlier, when, drowning in the Big Data tsunami, the social networking giant realized that it would need to "rethink its infrastructure to accommodate the huge influx of new people and data, and also control costs and energy consumption," the OCP Web site explains. The goal of the project was simple, but daunting: design the world's most energy efficient datacenter. Facebook assigned a small team of engineers to the task, and two years later its designed-and-built-from-the-ground-up datacenter project was up and running in Pineville, Ore.

Facebook open sourced the initiative and, along with its partners, formed the Open Compute Project Foundation, which functions as a standards body similar to the Java Community Process and the Eclipse Foundation. There's a board of directors to run the organization, a mostly elected incubation committee to approve new specifications, and project leads heading up the various OCP projects. Jason Taylor, vice president of infrastructure at Facebook, is the current president and chairman of the OCP Foundation board. His term ends in until October of this year.

Microsoft joined the OCP in 2014, announcing that it would be contributing specs and designs for the cloud servers powering Bing, Windows Azure and Office 365. Apple joined the community in 2015, as did HP, Cisco and Juniper, among others. Amazon Web Services remains conspicuously absent from the community, which seems to have collectively decided that there's no long any need for multiple datacenter hardware specs. Google joined the OCP last year.

"As our infrastructure has scaled, we've had to continue to innovate," Murillo explained. "By designing and building our own servers, we've been able to break down the traditional computing components and rebuild them into modular disaggregated systems. This allows us to replace the hardware or software as soon as better technology becomes available, and provides us with the flexibility and efficiency we need. We're thrilled to bring this flexibility and efficiency to the industry through the Open Compute Project."