Based on CRUSH algorithm, Ceph divides and replicates data into different storages. In case one of the storages fails, the affacted data are identified automatically; a new replication is formed so that a required number of copies come into existence. The algorithm is defined by so called Replication Factor, which indicates how many times the data should be replicated. One of the biggest advantages is that this factor can be changed at runtime, Ceph will then automatically create the necessary copies for already stored data.
