i think this is all based on the VL2 paper, described here: http://perspectives.mvdirona.com/2009/10/05/VL2AScalableAndF... and the paper available here: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.156....

it boasts of some freakishly fast ability to shuffle 2.7Tb between 75 servers in 27 seconds...