Blog
Wednesday, 12 March 2008Implementing Sphinx search
In a recent project we were asked to implement a full text search engine called Sphinx. Sphinx was created back in 2001 because the creators could not find a decent search solution for a database driven website. Due to the really fast searching and indexing, Sphinx is currently used on a wide variety of sites which all have to handle a lot of data (think of sites as blogcatalog.org and thepiratebay.org).
As I have the most Linux experience of my team, I was asked to install and implement sphinx search on our development server. After the compiling (which went perfectly right out of the box), the difficult part started. Because nobody had any experience with sphinx (we usually employ Google search, Htdig or Zend Framework's Lucene engine), the configuration started a bit rough. But after a while (and a lot of changes and testing) it started to work really fast. During the implementation I really noticed the fast indexing Sphinx can perform. Depending on the data and cpu of the system it can index up to 10MB per second.
For the current project it fulfills our needs, but there is much more to Sphinx than what we currently worked with. I'm looking forward to work more with Sphinx and discover its full potential (including the Sphinx storage engine which can be compiled into mysql). More info on Sphinx can be found here: http://www.sphinxsearch.com/ Share         |
