Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

This question already has an answer here:

With millions of users searching for so many things on google, yahoo and so on. How can the server handle so many concurrent searches? I have no clue as to how they made it so scalable. Any insight into their architecture would be welcomed.

share|improve this question

marked as duplicate by Michael Hampton Oct 9 '13 at 2:40

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

2  
Considering how thousands of great Google engineers have been working on this for over a decade, it's probably fair to say a single ServerFault question won't do it any justice –  philfreo May 29 '10 at 17:22
add comment

3 Answers

There are numerous case studies and talks by Google engineers available online with a little searching. Suffice it to say that Google Search is highly distributed and pushed out datacenters all over the world.

There's a ton of information available over at http://highscalability.com/google-architecture.

share|improve this answer
    
One example is the paper published years ago on disk failures (at the time, estimates were they had 800,000 servers). And the paper on MapReduce, just to name a two. –  ChuckCottrill Oct 9 '13 at 2:32
add comment

As has been mentioned, the networks and architecture of large-scale websites is highly distributed across many data centers and tens of thousands of servers. If you're interested in how this works, I'd recommend a book called Scalable Internet Architectures which describes some of the concepts and theories behind scalable and distributed systems.

share|improve this answer
    
One estimate I read a few years ago said 800,000 servers. –  ChuckCottrill Oct 9 '13 at 2:33
add comment

Among other things, they use 100,000's of servers, and MapReduce. Massively parallel.

share|improve this answer
add comment