Design and Implementation of Multi-Version Disk Backup Data Merging Algorithm

Abstract

Multi-version data management in disk backup and recovery is to manage the temporal attribute of backuped data. It can support to retrieve timestamp (time slice) disk data according to different query type. Exiting multi-version data management algorithms have two shortcomings. First, they are inefficient in multi-time point data query and updating which are adopted by data backup and recovery usually. Second, they use centralized data indexes which are not suitable for backup data management. To overcome these limitations, Backup Data Merging (BDM) algorithm is proposed in this paper, which uses distributed storage structure according to disk data format. By range operation, BDM algorithm can generate timestamp (time slice) data index dynamically. By comparing with traditional algorithms, BDM algorithm achieves high performance in storage utilization and query efficiency.