During the 24th minute of the recent Hadoop Summit Video [1] Avery Ching talks about how Giraph is made scalable. Since, Hama and Giraph are based on the BSP model, it should be possible to scale Hama in a similar fashion.
Basically, at the end of each super step the BSP tasks sends some metrics to the master and the master further partitions the data in the most loaded BSP tasks and uses the the free available slot to process them. Thanks, Praveen [1] - http://www.youtube.com/watch?v=b5Qmz4zPj-M<http://www.youtube.com/watch?v=b5Qmz4zPj-M&feature=youtu.be&a>
