Tomer Kaftan

As of pull request #51 being merged, this exists. There is now room for some extra optimizations in terms of only shipping new observations.

As of pull request #44 this may be as simple as changing the spark data location config to be in the tachyon filesystem

As of Issue #51 Broadcast via the spark filesystem exists. In the future it would be nice to have broadcast via HTTP, and broadcast via Torrent