How it is possible to serve web services to entire planet with various adroit applications of search running on the front end and worthy database engines at the back end? Well, Cluster Architecture by Google provides an effective way.
Recently, I came over an IEEE paper of Cluster architecture from Google System Lab which elucidates all the enigmas of search engine architecture. This paper describes the way their (great Google) search engines works, a manner in which a search query ends up to appropriate results and the amount of computing required to carry out a search operation. Well, it’s quite hard to generate resultsets after a single query operation for such a huge database in just no time. On an average, a single query on Google reads hundreds of megabytes of data and consumes billions of CPU cycles. But, the cluster architecture comprises of such a heavy amount of parallelization that lets multiple queries run on multiple processers and to equilibrate the workload of search. The Google web server consists of various index servers and documents servers which plays there prominent role to serve as per the request.
Google Query serving architecture:
This architecture’s engages more than 15,000 commodity-class PC’s with fault tolerant software’s to deploy the workload of multiple queries, thereby achieving a superior performance at a fraction of cost of system built which is moreover a system built from fewer, but more expensive, high-end servers.
Nevertheless, using inexpensive PCs to handle Google’s large-scale computations has drastically increased the amount of computation that they can afford to spend per query, thus helping to improve the Internet search experience of millions of users.
This paper is really a fascinating stuff of knowledge about Web search servers; especially, the part of doc servers and Index servers. Kindly read this document following is the link of that.
GOOGLE WEB SEARCH: An IEEE white paper By Luiz André Barroso, Jeffrey Dean, Urs Hölzle
0 comments:
Post a Comment