Posted by: Peter Burrows on April 2, 2009
For years, Google hid details regarding the type of servers it uses in its data centers to deal with all those queries as though it were a government hiding launch codes for nuclear missiles. The approach not only kept copycats away, but added a dash of mystery that heightened the general belief that Google was light years ahead of the competition when it comes to operating these vast facilities.
But over the past year, that secrecy was getting a little old. In fact, it created an opening for Microsoft to position itself as a more benevolent counter-weight of sorts. When I wrote this story on Microsoft’s huge data center ambitions late last year, company executives argued that their company was just as good if not better than Google at running data centers—and was certainly better for the industry because it was willing to share details of its approach to help drive innovation and best practices among would-be suppliers.
Well, Google has now taken a big step towards erasing that marketing advantage for Redmond. As CNET’s Stephen Shankland documents in this piece, Google described its server architecture in great detail.
I haven’t done any reporting on the topic, but let’s just say it wasn’t a good day for Microsoft’s claim that it’s as good as its search nemesis. For example, Microsoft has been crowing about the fact that it opened up a huge new data center in Chicago in which all of the servers are stored in shipping containers, an approach that makes it simpler and cheaper to deploy hundreds of servers at once. Turns out Google has been doing this since 2005. And while Microsoft has a goal to achieve a “power usage effectiveness” score of 1.2 (where a score of 1 means all of the electricity consumed is used to run the computers, rather than to cool and light the facility) in the next few years, Google has already hit this level as of the fourth quarter of 2008. Ouch.