Open Compute Project- Last Year, when he appeared onstage at the Mobile World Congress in Barcelona, Mark Zuckerberg painted Facebook as a friend of the world’s wireless carriers-not a foe.
Sure, Facebook was building drones, satellites, and lasers as a way of bringing wireless service to all those people on Earth who don’t already have it, and that might seem like competition for the mobile establishment. But Zuckerberg, Facebook’s founder and CEO, insisted that the real leaders in this area were the existing operators, companies like Verizon and AT&T and Deutsche Telekom and Vodafone that have long delivered wireless service to our phones and tablets.
“It’s important not to lose sight of the fact that the real companies that are driving [the evolution of our mobile networks] are the operators and all the investments they’re putting together,” he pointedly proclaimed at MWC, where those operators gather each year to celebrate their own way of doing things.
For some, his words seemed disingenuous—maybe even a little audacious. After all, Facebook was also pushing various Internet apps that seemed to undermine those same mobile operators, including Facebook Messenger, WhatsApp, and Free Basics, its Internet service for people in the developing world. These apps offer things like free texting and free calling—in effect, services designed to replace services that generate an awful lot of revenue for the mobile operators. And indeed, some carriers saw it that way. (Free Basics in particular has angered operators as well as app builders and policy makers).
But twelve months on, with Zuckerberg due to appear at MWC for the second year running, Facebook has made good on his claims of mobile world friendship—at least in one very big way.
Today, Facebook unveiled a new project that seeks not only to accelerate the evolution of technologies that drive our mobile networks, but to freely share this work with the world’s telecoms. Working alongside Deutsche Telekom (the largest wireless carrier in Germany), SK Telecom (the largest wireless carrier in South Korea), and Nokia Networks (which supplies much of the network hardware used by the world’s carriers), Facebook plans on building everything from new wireless radios—the hardware that shuttles wireless signals to and from our phones—to new optical fiber equipment that can shuttle data between those radios. Then, the company says, it will “open source” the designs, so that any wireless carrier can use them.
The hope is that this will lead to better wireless networks—wireless networks that can keep up with all the stuff we’re doing on our cell phones, from listening to music and watching videos to, yes, diving into virtual reality. “These really immersive experiences are all looming,” says Facebook’s Jay Parikh. They’re looming not only for the telcos, but for, well, Facebook itself. That’s why the company is launching this new project. Facebook wants to ensure that the telcos can deliver all the video—and all the virtual reality—it will stream across its social network, all over the world, in the years to come.
Parikh, Facebook’s vice president of engineering, drove the creation of what he calls the Telecom Infra Project. That’s Facebook-speak for telecommunications infrastructure project, but you can just call it TIP. This big idea echoes another seminal Facebook effort: the Open Compute Project. With Open Compute, Facebook remade the hardware inside the massive computer data centers that drive modern Internet services and then freely shared the designs with anyone who wanted them—helping to upend the traditional server and networking industries in the process. With TIP, Facebook is poised to do much the same with the hardware that telecoms use outside the data center.
This might involve the drones and lasers Facebook is already building inside its own Connectivity Lab as it works to bring Internet access to the developing world, but TIP will also extend all the way to the edge of mobile networks—from the massive base stations that serve as hubs for the network to the wireless radios that transmit data to consumer smartphones from local towers.
Late last month, Facebook launched a new effort inside the Open Compute Project that seeks to help telecoms improve the hardware inside their data centers. Now, it also aims to help them improve the hardware across the rest of their networks—to help them expand and enhance their networks at a much faster rate. “The only path that I know that works is to basically take a couple of pages of our playbook for open source software and the hardware and data center work we’ve done, and try to approach the telecom infrastructure problem in a similar vein,” Parikh says.
For Axel Clauberg, a vice president of architecture at Deutsche Telekom, the project makes good sense—not just for Facebook but for the telecoms. “We believe that the exponential growth of Internet traffic requires new approaches,” he says. “The Open Compute Project has proven that open specifications for hardware, combined with an active community can have a drastic impact on efficiency and cost. TIP will trigger the same for all areas of the network.”
Erik Ekudden—a tech strategist at Ericsson, which, like Nokia, builds much of the gear that telcos used outside the data center—also sees potential in this fundamental idea. Lessons that companies like Facebook have learned in the data center, he says, could help telcoms improve their mobile networks. But he also says that ideas can move the other way—from the telcoms to Internet giants like Facebook.
A Better Network for Facebook
Does it really make sense that Facebook, of all companies, is running this project? Well, the company has long worked to build new telecom technology under the aegis of its Connectivity Lab. And after years spent running the Open Compute Project, it certainly understands how to push hardware markets in new directions.
Over the last several years, as Facebook expanded its online empire to hundreds of millions of people, company engineers came to realize they needed a new way of building this empire. They couldn’t use traditional hardware from traditional suppliers like Dell, HP, and Cisco. It was too expensive, too elaborate, and too difficult to operate at such an enormous scale.
So, they built a cheaper, more streamlined, and more malleable breed of gear, including computer servers, data storage devices, and networking switches. The basic idea was to build very large networks from very small and very cheap pieces that could be easily reprogrammed—and easily replaced.
In building a similarly large empire, Google had also designed its own hardware. But Facebook went further. Through the Open Compute Project, it open sourced this gear, and it encouraged others to do the same.
The result was an entirely new market for data center hardware. Big online operations—like Apple and Rackspace and Fidelity—suddenly had access to hardware designs that better suited their needs, and with many hardware vendors embracing these designs and turning them into commercial gear, they had more choice. Tech giants like Apple and Rackspace and Microsoft have all joined the Open Compute Project, as have finance giants like Fidelity, Goldman Sachs, and Bank of America—all looking to streamline and improve their data center hardware in similar ways.
About two years ago, during a meeting with Zuckerberg at Facebook headquarters, Parikh suggested the company do the same thing with telecom hardware. Facebook wanted the telecoms to improve their networking infrastructure at a much faster rate, and Parikh saw an obvious way of encouraging this speedup. “We had this Open Compute thing,” Parikh remembers, “and it dawned on me that the same kind of problem is facing the telecos—but it’s probably twenty times worse. There’s a lot less choice there. These networks are expensive. They’re hard to deploy. They’re hard to operate.”
Roll Your Own Drone
But the idea needed a few years to gestate. In March 2014, Facebook launched its Connectivity Lab. Led by Yael Maguire, who had previously driven much of the company’s Open Compute work, the lab aimed to build a wide range of technologies that could deliver Internet access to the hinterlands around the world. Following similar work at Google, this effort included satellites that would circle the Earth and drones that would float through the stratosphere as well as a new breed of laser that could move data between such hubs.
When we spoke to Parikh about a year into this effort, we asked if the Open Compute Project could serve as a model for Facebook’s work with drones and satellites and lasers. “Absolutely,” he said. “A lot of the things that we’ve learned—that we’ve been working on from a philosophical perspective in [the Open Compute Project]—help us get in the right mindset when it comes to connecting four to five billion more people to the Internet.” He stopped short of saying the company might open source its drone designs. But now, in launching the Telecom Infra Project, that’s exactly the kind of thing the company is looking to do.
In other words, anything that the company builds inside the Connectivity Lab could trickle down to TIP, which means it could be open sourced. That might include drones and lasers as well as other gear. “The only way that this [Connectivity Lab] technology has any impact,” says Maguire, “is if we’re open about it.”
At the same time, TIP will explore technology that can push places like the US and Europe towards “5G” technologies—which would offer speeds well beyond today’s 4G networks. “It shouldn’t take ten years to upgrade your network to get to that new technology,” Parikh says. And in the developing world, he hopes to move networks beyond the much the slower 2G tech that are now the norm.
In particular, Parikh says, the project could explore new radio technologies along the lines of the pCell radios developed by serial inventor Steve Perlman and his company, Artemis Research. pCell antennas don’t just blanket a neighborhood with a single signal, or cell, that all phones and tablets share. They broadcast signals that combine to create a “personal cell” that follows your phone from place to place. Perlman has estimated that this technology could increase wireless speeds by 1,000 fold.
So far, Perlman has had relatively little luck pushing his tech into big-name telecoms. But Nokia Networks, one the major hardware vendors that supplies the telecoms, is kicking the tires on pCell—and Nokia is now part of TIP. The project includes not only Facebook and various telecoms, but companies that already build telecom hardware. Chip maker Intel is also a founding member.
The politics of a project like this can be delicate. The other big telecom suppliers—Ericsson and Huawei—have not joined the project, and they may take issue with the idea of open sourcing designs. After all, they make their money from proprietary gear. And, well, not everyone wants Facebook sticking its nose in their businesses. The company has received an enormous about of criticism over Free Basics, with many saying that the company shouldn’t naturally assume that other parts of the world can’t take care of themselves. But, well, this is how Facebook operates: with a will to make big changes, and the resources to follow through. With projects like Open Compute—-which, unlike Free Basics, is a long way from the buzzsaw of the net neutrality debate—this attitude was enormously effective.
“We are expecting similar changes [to what] already happened within the data center,” says Deutsche Telekom’s Clauberg. “While this impacts existing players, it also creates large opportunities.” He means opportunities for telcoms and the companies that build hardware for telcoms. But the project will also bring new opportunities for the rest of us. If wireless networks get better, we, the consumers, are the ultimate beneficiaries. We’re the ones watching all that video—and all that virtual reality. And spending all that time on Facebook.