An adaptive selection method for repair nodes in distributed storage systems
Rong Chang, Chuanxu Yang, Yuankang Lei · Journal of Physics Conference Series · 2021
Abstract In distributed storage system, how to optimize the repair time of failure data to ensure the high reliability of the system has aroused wide attention. In recent years, studies have found that different node selection mechanisms in the repair process have a great impact on the data regeneration time, and previous work has proposed the node selection under the single node failure scenario. In the case that multiple nodes are repaired at the same time in the system, the huge space-time overhead of SPSN algorithm makes the data regeneration time no longer optimal. Make statistics on failure data and causes of existing real systems; In order to solve the replication strategy in the storage overhead and the erasure code strategy in the repair of bandwidth overhead, network coding technology, called regenerative code, is introduced into distributed storage system to balance the storage cost and bandwidth cost. In this paper, single-node repair algorithm and multi-node collaboration algorithm are mainly studied.