Reconfigurable distributed storage in "cloud"

Distributed storage is the basis of the "cloud computing". It replaces the replicas in different locations. Replication had been the main method to improve the reliability of cloud computing. It became the base issue that how to maintain the consistency between replicas. This paper imposes distributed share memory algorithm which is based on primary-backup, it can dynamic reconfigurable replicas, so that improving availability and reliability of the system, and maintain consistent between replicas. First, the system sets the auxiliary master, the master storage the configuration of the replicas and active replicas. The auxiliary master imposes configurable request, but read/write request is imposed by clients; second, read is optimized by reading one, write operation can complete when some replicas failed; third, read/write operation can concurrently run with reconfigurable operation; final, the algorithm can tolerance n-1 failure nodes.