12 April, 2011

Google’s secret sauce

Google's secret sauce - or rather one of them - is its operational expertise at deploying and managing vast (I mean really vast) numbers of servers in its operation.

Google’s Servers
It uses commodity computers with its own Linux-based Operating System (see Wikipedia), both of which have been tuned to Google’s requirements.

It is ‘reasonably’ estimated that it has 1 million of them - the largest server farm in the world by far. Here’s an eye watering graphic from Gizmodo: Google’s Insane Number of Servers Visualized (April 2010).

(Want to know who the other big server boys? see DataCenterKnowledge)

Google are notoriously secret about its hardware, so I think these figures are only ball park estimates.


Performance IS important
This operational horse power is a key competitive advantage is Google's business operation. For example, user responsiveness is critical. These stats are taken from the Velocity Conference 2009 and reported by its organiser, Steve Souders on his blog post, Velocity and the Bottom Line:
  • Bing found that a 2 second slowdown changed queries/user by -1.8% and revenue/user by -4.3%.
  • Google Search found that a 400 millisecond delay resulted in a -0.59% change in searches/user.
  • At Google, one experiment increased the number of search results per page from 10 to 30, with a corresponding increase in page load times from 400 milliseconds to 900 milliseconds. This resulted in a 25% dropoff in first result page searches.
  • At Shopzilla, a year-long performance redesign resulted in a 5 second speed up (from ~7 seconds to ~2 seconds). This resulted in a 25% increase in page views, a 7-12% increase in revenue, and a 50% reduction in hardware.

Google’s own dedicated network
Google itself has a massive internal network. If you think of all the crawling that Google does, then the spider will be want to shuttle that information back to its nearest home base. Given the number of sites that Google crawls, it’s a really good idea to have ‘home’ nearby.

Here’s one estimate for the volume of traffic:
60 percent of Google's traffic was being channelled through direct interconnects that link its massive data centres to one another.
Significance?
Given Google's secrecy around its operational expertise, then it should take note of Facebook open sourcing its data centre design recently.

No comments: