Controlled Replication Under Scalable Hashing (CRUSH) is an algorithm used in Ceph to store and retrieve data by computing data storage locations within the cluster. It does so by providing a per-device weight value to distribute data objects among storage devices. The value is auto assigned, based on the actual size of the disk drive being used. For example, a 2 TB disk drive may have an approximate weight of 1.81. The drive will keep writing data until it reaches this weight. By design, CRUSH distributes data evenly among weighted devices to maintain a balanced utilization of storage and device bandwidth resources. A CRUSH map can be customized by a user to fit any cluster environment of any size.
For more details on CRUSH maps, refer to the following link: