Google’s Architecture Described

Posted by Sam Churchill on

Speaking at the Open Networking Summit 2012, Google detailed their massive global backbone (pdf). Google’s ambition is to build a WAN that is higher performance, more fault tolerant and cheaper.

Google is now using OpenFlow as the basis for its so-called G-Scale network that links its global data centers. OpenFlow is a technique for controlling network operations in software, run on centralized computer servers saving cost, time and power.

It aims to simplify and virtualize today’s business networks. So far it has found it can run such functions “literally 25 to 50 times faster on a 32-core workstation,” said Urs Holzle, senior vice president of technology infrastructure at Google. Google typically treats its infrastructure as a trade secret best kept hidden from competitors. Hölzle’s speech provided a rare window into how a portion of the company’s operation works.

Like fellow cloud data giants Google, Amazon, and Microsoft, Facebook is designing and building its own data centers. Facebook is sharing its data center design through the Open Compute Project. The company’s first custom-built facility went live in Prineville, Oregon at the beginning of 2011. Apple bought 160 acres in Prineville near the Facebook data center and is now constructing a 10,000 square-foot, server farm. Apple’s 31-megawatt data center in Prineville will use enough energy to power five cities of Prineville’s size.

Posted by Sam Churchill on Friday, April 20th, 2012 at 7:36 am .

Leave a Reply