Utilizing CRUSH Algorithm on Ceph to Build a Cluster of Reliable Data Storage

Ceph uses CRUSH algorithm (Controlled Replication Under Scalable Hashing) to achieve data redundancy in storage cluster. The use of CRUSH algorithm can replace the role of controller to achieve data redundancy, so that dependency on vendors can be eliminated. CRUSH algorithm allows one to build a data storage cluster that has high scalability and flexibility. Unfortunately, no study has discussed comprehensively how Ceph uses CRUSH algorithm to create redundant data and spread it to storage clusters. This paper aims to examine the mechanism for placing Ceph data using CRUSH algorithm comprehensively. The research method used is by conducting literature studies and experiments. The experiment was carried out by directly trying out the configuration applied to the data storage cluster. By understanding the workings of CRUSH algorithm, it is expected that people can customize the data placement carried out by Ceph according to the needs of services that will be run on the data storage cluster.