基于遗传算法的副本管理策略  

The strategy of Replica Management Based on Genetic Algorithm

作　　者：张新亮[1] 万晓冬[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016

出　　处：《计算机仿真》2009年第1期197-199,263,共4页Computer Simulation

摘　　要：在数据网格环境中,为了提高数据的可靠性和降低用户对数据访问的延迟,广泛采用了数据复制技术;由于副本技术的引入,就会面临一个很直接的问题,即一个应用程序如何根据副本性能和访问特性,从一组副本中选择一个最佳副本,就是副本选择问题。针对数据复制技术中的关键技术:副本创建和副本选择,在价格机制模型和并行数据传输的基础上,采用了遗传算法实现副本的优化管理。最后使用网格模拟器OptorS im对算法进行测试分析,结果显示基于遗传算法的策略有更好的性能。Replication technologies have been widely used in data grid in order to improve data reliability and reduce the data access latency. As a result of the replica technology＇s introduction, a problem emerges directly, namely how can an application procedure choose the best replica from group of replicas according to the transcription performance and the visit characteristic. For the key technologies in data replication technology： the replica generation and replica selection, a genetic algorithm is used for optimal management of replica based on the price mechanism and parallel data transmission. Finally, the grid simulator OptorSim is used to test the algorithm, and the result thows that the strategy based on genetic algorithm has better performance.

关 键 词：副本管理 遗传算法 数据网格 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]