Provide a mechanism to copy a CF from the production cluster to a development cluster

[Evan]

I would like to work on remodelling the DayBucketsCount Column Family with an emphasis on read performance. To measure this accurately, it would be ideal if I could work with a copy of the real data. Cassandra has a few approaches available for copying this out:

- sstable2json
- nodetool snapshot
- describe keyspace | ssh host cqlsh, rsync, and nodetool refresh
- sstableloader

Do note that when we're dealing with sstables, we'll need to operate on all nodes using a tool like pssh, as the keys for different rows will exist on different nodes.

See also:
http://www.grulic.org.ar/~mdione/glob/posts/restoring-cassandra-online/
http://www.datastax.com/docs/1.1/operations/backup_restore
http://www.datastax.com/dev/blog/bulk-loading