摘要
在分级存储中,业务数据的访问模式多样且多变,使得为其选择合适的数据迁移策略非常困难,为此引入迁移策略优先级的概念:系统内的迁移策略按优先级高低排序,每次启动数据迁移时,都选择当前优先级最高的策略作为迁移所用策略,并且在数据迁移进行时以及完成后对策略的优先级进行动态调整,从而达到系统自适应选择最优迁移策略的目的。原型实验证明,该方法确实能够根据优先级自适应地选择出最佳策略,从而降低数据迁移对前台应用的影响。
In the hierarchical storage management （HSM） ,it is difficult to choose a data migration policy which meets precisely the diverse and the changes in business data access patterns, so that the concept of policy priority could help to quickly choose the most appropriate policy for the HSM system. The value of a policy priority is dynamically adjusted before and after a migration which is driven by the policy, and then, the HSM system chooses the policy that has highest priority value as the policy for the next migration. After a few adjustments like this, the system will obtain the optimal policy. The prototype experiments show that the method can indeed adaptively choose the best policy by dynamically adjusting the priority, therefore,the impact on the business applications is greatly reduced.
出处
《计算机科学》
CSCD
北大核心
2012年第3期139-143,共5页
Computer Science
基金
国家863重大项目(2009AA01A404)资助
