This is the algorithm used in Hadoop's record setting TeraSort benchmark:

http://perspectives.mvdirona.com/2008/07/08/HadoopWinsTeraSo...

http://svn.apache.org/viewvc/hadoop/core/trunk/src/examples/...