Study On The Repair Of Failure Node In Distributed Storage Systems

In recent years,with the rapid development of Internet and Information Technology,the quantity of data is showing explosive growth.All kinds of Internet applications and the scientific researches have generated huge amounts of data.In order to meet the need for mass data storage and processing,distributed storage technology has been proposed and widely used.In the view of universality of failure nodes,it is very important for a failure node to be repaired quickly and effectively to maintain the reliability of the system.In this article,we consider a single and multi-node failure,and the main research is as follows:(1)In order to speed up the failure repair rate of single node in storage system,and reduce the frequency of other nodes' failure when repairing the failure nodes,for maintaining a small storage overhead,we introduce the Similar Local Reconstruction Code(SLRC).According to the Local Reconstruction Code encoding scheme with the local check block thought,this paper further reduce the single node repair bandwidth overhead,and decrease the computational complexity of the repair of failure nodes.The encoding scheme increases some storage slightly,but improves much repairing performance of single node failure and system scalability.(2)Regarding the repair of multiply nodes failure problems in Distributed Storage System,this paper proposes a Locally Repair Code scheme based on Pyramid code.The coding scheme adopts the minimum achievable coding structure of Pyramid code,then divides it into some local repair groups,and ensures low repair locality and realizes to repair the failure node effectively.The analysis shows that the Locally Repair Code based on Pyramid code can repair multiple fault node rapidly in storage systems,In addition,it has low repair locality,and is more superior to the three-copy replication strategy and the simple Regeneration code.Last but not least,the Locally Repair Code based on Pyramid code has relatively small storage overhead and repair bandwidth overhead.