Deletes in the world of distributed databases are hard to do. After all, how do you replicate nothing? Especially a key for nothing which once held a value. Tombstones are Cassandra's way of solving that problem.
As Cassandra uses a log-based storage engine, deletes are essentially writes. When data is deleted, a structure known as a tombstone is written. This tombstone exists so that it can be replicated to the other nodes that contained the deleted replica.
Discussion about tombstones tends to be more prevalent in Cassandra application-developer circles. This is primarily because their accumulation in large numbers can be problematic. Let's say that data is written for a specific key, then deleted, and then written and deleted ...
