海量数据查询分析系统备份与恢复策略设计

The Policy of Backup and Restore for the System of Mass Data Querying and Analysis

解决海量数据查询分析系统存在的数据量与查询速度,数据访问局部性与数据无限制性需求间的矛盾,提出一种备份和恢复策略,备份加载前的数据,并将业务数据的时间属性与备份任务关联,与基于DBMS的备份和恢复方式相比,降低了海量数据对磁盘阵列空间和备份操作对系统资源的占用,增加了恢复数据的可操作粒度和配置灵活性。

The system, querying and analyzing massdata for DSS, usually demands abundant data, which occupy over much disk space and result into augmented costs and complexity for managing. While the data frequently accessed is only partial. To solve above problem by dividing data into two parts, the latest that will be frequently accessed stores in RAID and the far that will be seldom queried places on tape . The subset that is stored on tap would be moved to RAID from tape when it will be accessed. This paper describes a policy mainly for moving data between RAID and tape, which comes from fault - tolerance and disaster - recovery. Store the data on tape before itloaded into database and link its time attribute withbackup task in log, which would be visited when restoring.