Facebook’s Open Compute Shapes the Future of Big Data Servers

When a company grows to the size that Facebook has, vendors no longer get to dictate the specs of its server acquisitions. Instead, the boot is on the other foot Facebook gets to determine exactly what it wants, with customized server requirements made especially to order to meet its unique and over-sized needs.

Now, Facebook is hoping to take things one step further with a proposed new model for designer servers that it claims will give enterprises more flexibility when it comes to upgrading their servers and adding new components.

The social media giant says that it wants to “disaggregate” date centers, reducing the dependency on specific, branded computer parts and instead, make it easier for businesses to customize server designs to meet their specific workloads. Facebook laid out grand plans for its scheme during Wednesday’s Open Compute Project summit, revealing how it’s working alongside component suppliers and enterprises to design the next-generation of data center parts.

Introducing the Next-Generation of Data Servers

 

So what is Facebook up to exactly? Essentially, through its Open Compute Project, it’s looking for ways in to reduce the hassle and costs associated with the hardware that underpins today’s data centers. Wired.com points out that the project was founded by Facebook under the leadership of Frank Frankovsky, and is urging an unheard of level of cooperation across the data industry, pressing vendors to collaborate on new hardware designs. The group – now a non-for-profit operation – has attracted the support of numerous companies, including Rackspace, Fidelity, AMD, Dell and Intel.

Each of these companies is coming up with its own ideas on how the next-generation of data center hardware can be built, but what Facebook announced yesterday could well prove to be revolutionary. The social media company has come up with a unique, consistent slot design for motherboards that’s compatible with chips from any vendor. Up until now, it hasn’t been possible to mix and match different chips – for example, if a business wanted to swap its Intel chips for AMD chips, it would need to rebuild its servers to make them compatible.

Frankovsky calls his new invention the “group hug” slot. Not only can it accept silicon chips from any vendor, but the design has already been welcomed by the likes of Intel, AMD, Calxeda and Applied Macros, all of whom have committed themselves to supporting it.

Facebook says that the design will help to resolve issues wherein customers that buy one server set up find themselves limited to choosing components only from the vendor which supplied the system in the first place.

“The power supply shouldn’t be embedded in the server, otherwise you have to design the server around it, and if your power requirements change, you’re stuck,” goes the logic of Frankovsky.

It’s not only Facebook that’s being innovative here. The chief vendors are also keen to weigh in with their own ideas. Chief among these is Intel’s plan to introduce its super-fast silicon nanophotonics networking connection based on fiber optics that will speed up the transfer of data between chips on a rack. Intel plans to announce this later this year, with Facebook likely to be one of the first customers to get their hands on the new tech.

Silicon photonics will do a lot more than just speeding up communication between individual chips. GigaOM’s Stacy Higginbotham points out that by using the right hardware, businesses can run a rack directly into a server, an implementation that would render top-of-rack switches irrelevant.

Meanwhile AMD has just announced the first of its own offerings to the Open Compute community, with the release of a new motherboard designed to run inside specialized servers purpose-built for specific workloads. Codenamed Roadrunner, the Open 3.0 chipset combines speedy processes with design methodologies developed by one of the biggest web-scale companies around. It also supports both standard 19” rack environments and the Open Rack specification, a hardware standard developed by Facebook for the purpose of reducing its data centers’ footprint.

For big web-based businesses, these developments are all great news. More flexible servers will cut out waste and allow IT teams to transform their data hardware assets into something they can upgrade at will, easily and at a less prohibitive cost.

About Mike Wheatley

Mike Wheatley is a senior staff writer at SiliconANGLE. He loves to write about Big Data and the Internet of Things, and explore how these technologies are evolving within the enterprise and helping businesses to become more agile. Before joining SiliconANGLE, Mike was an editor at Argophilia Travel News, an occassional contributer to The Epoch Times, and has also dabbled in SEO and social media marketing. He usually bases himself in Bangkok, Thailand, though he can often be found roaming through the jungles or chilling on a beach. Got a news story or tip? Email Mike@SiliconANGLE.com.