So what’s a company to do when faced with the daunting digital task of hosting some 140 billion photographs and counting? If you’re Facebook, and you’ve recently filed papers for perhaps the largest initial public offering in history (upwards of $5 billion), how about totally re-imagining the way servers and storage hardware are built?
In an article published late last week, Wired magazine contributing writer and editor of Wired Enterprise, Cade Metz, profiled recent developments happening behind the scenes at the Menlo Park, California-based social networking giant. Chief among them, Facebook’s ongoing plans to expand upon its Open Compute Project, the 2011-launched initiative with a simple mission: “build one of the most efficient computing infrastructures at the lowest possible cost.”
Here at Chatsworth Products, Inc. (CPI), we're proud to count ourselves among the Open Compute Project ranks, having taken part in past summits and making plans to further engage in forums dedicated specifically around optimized power and cabinet infrastructure.
And when more of the world’s most forward-thinking high-tech experts assemble in May at the next Open Compute Summit, you can be sure that even the most radical ideas around data center efficiency and operations will be on the table for discussion.
That includes some choice insight reported on by Metz and offered up by the man who spearheads hardware design at Facebook, Frank Frankovsky.
“Wouldn’t it be cool if you could actually disaggregate the CPUs from the DRAM complex?”
This, Frankovsky says to Metz, is the byproduct of an ambitious idea being considered, whereby a server’s physical infrastructure is essentially split apart into separately housed core components. Think of it as increased efficiency (easier access to parts for maintenance, upgrade, etc.) by way of decreased consolidation (who says everything under one roof is the only way to live?).
Perhaps most intriguingly, it’s what happens after such a concept is conceived that interests us most here at Chatsworth Products, Inc. (CPI). Metz explains the possible outcome of a new approach to this kind of server design as follows:
“Project members are also working to create a new rack design that can accommodate this sort of re-imagined server infrastructure. A traditional server rack houses several individual machines, each with its own chassis. But the Open Rack project seeks to do away with the server chassis entirely and turn the rack into the chassis.”
As experts on rack design and manufacturing, you can be sure that CPI will continue our participation in the Open Compute Project, as we all work collectively towards optimized infrastructure.
Ultimately, it is this environment of collaboration that truly represents the concept of the Open Compute Project at its core. Engage the industry at large, promote innovation, map its potential, act on what works, and share with the world.
In fact, it's a concept we've covered before, back when Facebook's Prineville, Oregon data center made its debut with some outside-the-box approaches to data center cooling.
But speaking of sharing, there’s something I’ve been wondering – 140 billion pictures multiplied by the notion that every picture is worth a thousand words equals… oh forget it! I’ve “racked” my brain enough for one article!