Thursday, July 13, 2006

The technology behind Google

"How Google Works" is a summary of the interesting technology used by Google. Among the highlights:

  • Built their own highly distributed Google File System (GFS). This is really an application storage utility rather than integrated with the operating system, and "relaxes" consistency for performance. (See "The Anatomy of a Large-Scale Hypertextual Web Search Engine" and BigTable.)
  • The secret PageRank algorithm.
  • Pre-packaged racks of cheap computers that can be deployed over night at a data centre near you.
  • Go for server hardware that uses less power (rumor that Google switched to AMD just because their processors ran cooler) so you can pack more servers into a datacenter.
  • Developed their own project management software that allows users to enter a summary of what they did this week and what they plan to do next week. Nothing special there except the software analyzes free-form text instead of forcing the user to enter text in a structured, less flexible way (e.g. an HTML forms application).

No comments: