Cassandra storage utilization for RESTBase is getting quite high; At the time of this writing, the cluster-wide average is ~60%.
Past efforts to cull revisions were meant to delete everything older than December 2015, yet there is evidence that a not insignificant number of records still exist. Since it should be uncontroversial to delete everything older than December 2015, I propose we treat this as the lowest hanging fruit and begin there. And, as it will take some time for tombstones to work their way through the compaction pipeline, I propose we start this as soon as possible.