Frank Frankovsky, Facebook’s vice president of hardware design and supply chain optimization, who helped oversee the development and growth of the company’s custom server effort, has left the social networking company to form his own as-yet-unnamed startup that will focus on building optical storage for the enterprise.
In an interview with me, Frankovsky said he had resigned from Facebook last week to pursue this idea. Meanwhile, Jason Taylor, Facebook’s director of infrastructure, has assumed responsibility for the hardware design and supply chain teams at Facebook and will continue working with the Open Compute Project on Facebook’s behalf.
Taylor has been overseeing much of that work for the last year, according to Facebook, and he will also be joining the Open Compute Foundation board along with Bill Laing, corporate VP of Cloud and Enterprise at Microsoft. This brings the OCP Foundation board from five to seven participants. Frankovsky, who will remain chairman and president of the OCP board, will stay as an independent member.
Frankovsky joined Facebook from Dell in 2009 as the social network was thinking about how to rebuild servers for the webscale world. His work would later become the Open Compute Project, which was launched in 2011, and has since saved Facebook an estimated $1.2 billion in costs. Because the hardware developments it built and encouraged were open source and freely available, it’s also responsible for a lot of heartburn in the hardware business.
“When I joined, it was it was a culture shock,” said Frankovsky. “Being a hardware guy at such a fast-moving software company. It was like if you’ve ever jumped out of a fast-moving boat, you get the wind knocked out of you when you hit the water. That’s what it felt like to be a slow-moving hardware guy with all of these software people who were used to iterating every few weeks. That’s why we started Open Compute so we could speed up the development of hardware.”
Frankovsky, who was originally a business major who just happened into the computer trade thanks to an internship at Compaq during college, has been splitting his time between a home in Austin and Facebook’s Menlo Park headquarters. His new opportunity is a chance to spend less time on a plane flying between the two cities.
For his next career move he’s planning to build a hardware company that is reconfigured to work in a world where Facebook had commoditized much of the computing infrastructure with Open Compute. To do this, Frankovsky is thinking less about the hardware enclosures and more about what’s inside. In this case, it’s the storage medium and software to operate the storage.
The idea is to build new types of optical storage that can offer both more storage (more gigabytes!) and perhaps faster throughput (faster reads and/or writes). Today’s optical drives are familiar to consumers as Blu-Ray discs or CDs that don’t store much data — about 100 gigabytes per platter on the high end — making them impractical for enterprise use.
However, Frankovsky sees an opportunity. Today companies like Facebook deal with two extremes of storage — one is fast access to information and data to keep the site loading at a snappy pace, while the other is finding the cheapest place you can store those selfies from 2011 that you probably won’t want to access until you’re drunkenly scrolling through your timeline wondering where your youth went.
Optical storage could find a home in the second tier if it can be made more dense and cheaper. One way to do this is to enclose hundreds of platters containing a disc inside a rack and let a robot manage the loading of the platters. Facebook has already shown this off earlier this year with a 10,000 Blu-ray disc array. Facebook’s head of engineering Jay Parikh said in January that the system was capable of storing a petabyte of data while cutting 50 percent in costs and 80 percent in energy usage over its existing hard-disk-based cold storage methods. It will also provide 50 years worth of durability for data stored on the discs.
But this isn’t just a Facebook or webscale problem. If Frankovsky can bring disc storage in line with tape costs — a big if — or convince people that it’s much more reliable, there’s a large market of established businesses from financial firms to lawyers that keep data on tape.
Frankovsky’s idea is to implement a razor blade model for the storage media, with the storage media representing the blades that are upgraded and replaces over time. While Panasonic and Sony have a roadmap to bring 1 terabyte drives to production by 2016, Frankovsky believes there’s a lot of promise in boosting both density and throughput (he thinks he can boost it to 10 times the current optical disc rate) to be had in this area.
As someone who has sold hardware and bought it, Frankovsky certainly understands both sides of the equation. And since Open Compute has disrupted the hardware vendor market, it will be fun to see how one might go about building the next generation of hardware companies in this world with fewer customers and slimmer margins.