一种基于日志合并优化的数据同步机制  

A Data Synchronization Mechanism Based on Log-Merging Optimization

作　　者：张虎[1] 董小社[1] 伍卫国[1] 钱德沛[1,2] 戴罗庚[1] 

机构地区：[1]西安交通大学计算机系,陕西西安710049 [2]北京航空航天大学计算机学院,北京100083

出　　处：《小型微型计算机系统》2006年第12期2183-2188,共6页Journal of Chinese Computer Systems

基　　金：国家"八六三"计划项目"新型网络服务器系统(一)"(2004AA111120)资助

摘　　要：基于复制技术的高可用存储系统,在故障恢复时需要进行数据同步操作以保证数据副本之间的一致,缩减数据同步操作的时间可减少系统修复时间,从而提高存储系统可用性.数据同步过程的时间由数据同步所需传输的数据量决定.本文提出的基于日志操作合并优化的数据同步机制充分考虑到文件访问中的时间和空间局部性,通过对故障期间的所记录的日志进行合并优化,合并和去除不必要的日志项,使得需要同步的数据量最少,从而减少数据同步过程的时间.试验的结果表明,相对于其它几种通用的方式,基于日志合并优化的数据同步机制所传输的数据量最少,可有效的减少数据同步操作的时间.For a high available storage system based on replication technology, the data of two mirrored sites must be synchronized while the system is being repaired after one of data replica crashed. Accelerating data synchronization process can cut down the whole time cost of system recovery, and therefore improves the availability of the storage system. The time cost of data synchronization much depends upon the quantity of the data which is moved in the process. In this paper,a new synchronization mechanism is present to reduce such quantity to a maximum extent. This mechanism based on log-merging optimization,which merges and deletes unnecessary file operations recorded in the log file before data being moved,consequently,only the modified data will be synchronized. And the result of experiment shows that ,this mechanism transfers the least quantity of data than other widely used mechanisms,As a result,it takes least time to complete the synchronization and then enhances the availability of whole system.

关 键 词：复制 数据同步 日志合并优化 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]