Design And Implementation Of Backup And Recovery Tool For Distributed File Storage System

With the rapid development of Internet, the amount of data has increased enormously; the storage capacity of single PC could no longer meet the large scale file storage requirement. Thus distributed storage has emerged and grown rapidly. Since the system may be destroyed by natural disasters, sudden failures, it is necessary to backup and protect data in order to recover quickly if data has been destroyed.Based on JTang distributed file service platform developed by the lab, this thesis discusses problems concerning backup and recovery of large amounts of data under distributed storage environment including researches on efficiency, consistency, and reliability issues of the backup process in addition to fast recovery issues.On performance issues, taking advantage of file redundancy in distributed storage systems, a multi-source backup model is proposed. Under this model, backup files can be downloaded from several storage nodes considering issues like file size, load balance. As for consistency issues, snapshot technique has been adopted to get system's snapshot quickly, then do the backup. Incremental and differential backup are also supported by the system. To address reliability issues, resumable backup and its corresponding recovery algorithm is implemented to ensure the continuity of backup process in case unexpected errors occur.On data recovery, to improve efficiency of the recovery process, a multi-process model has been adopted. To ensure load balancing in the system, several nodes can be selected to recover data. And to facilitate management of backup and recovery processes, indexes have been built for the backup process to record information of each backup process and related information of files after the backup. Models have been built based on the backup process to facilitate system recovery and to manage the recovery process.Adopting models proposed in this paper and based on JTang distributed file service platform, high efficiency, reliability and recovery consistency have been realized in backup and recovery process. System testing data illustrates that the system can backup files efficiently while allowing users to access files normally. Moreover, the recovery and backup process can be resumed if being interrupted. At the same time when system failure occurs, the system can be recovered quickly.