A Faster Algorithm For Indexing - FastBit

Apologies, the marking of Challenge 17 has been a bit delayed but will be completed soon. Until recently I didn't know that the US Government (Well Lawrence Berkeley National Laboratory) hosts a Sourceforge type website called Codeforge. It uses the gforge.net software which is for non-commercial use and hosts projects from the laboratory. The most popular project there (by downloads) is an implementation of the fastbit indexing algorithm in C++. It's highly optimized for terabyte sized data searches using various techniques such as word aligned compression and in tests has outperformed MySQL by up to 100x though they quote 30x-50x as typical.

This remarkable performance has seen it win an award by R&D magazine in the 100 most innovative products (one of 4 from Lawrence Berkeley National Lab). The source code is still in beta but comes with make files, examples and tests for Windows as well as other platforms. There's a (pdf) article describing comparisons with MySQL.

• Link to C++ Code Library