一种基于拟序的关联备份任务调度算法  

A New Scheduling Algorithm Based on Quasi Ordering for Related Backup Task

作　　者：郑炜[1] 胡正国[1] 汪芳[1] 

机构地区：[1]西北工业大学计算机学院,陕西西安710072

出　　处：《西北工业大学学报》2004年第4期408-412,共5页Journal of Northwestern Polytechnical University

基　　金：西北工业大学青年教师创新基金 (1 0 0 0 M0 1 62 0 1 )资助

摘　　要：为了减少存储备份任务的执行时间和提高整个存储系统的效率,提出了一种新的SAN体系结构——基于集群技术的SAN体系结构,并在其结构上提出并实现了关联备份任务的数学模型和实现该模型的核心调度算法。该算法的基本思想是对投入的关联备份任务找到一个最优的执行顺序,以便提高整个系统的性能。实验数据表明,对于关联备份任务的调度与执行,该算法可以有效提高整个系统的效率。In order to reduce the execution time for storage backup tasks and improve the performance for the whole storage system, we present a new SAN(Storage Area Networks) architecture by introducing the cluster technique into traditional SAN architecture, building a mathematical model of related backup task and describing its kernel algorithm. Two kinds of transparency--transparency provides the user with much appreciated convenience--can be fulfilled by our system. Both the concrete execution scheduler and the concrete storage device are transparent to the users. Firstly, we present the architecture of the SAN system based on clustering technology; secondly, we design a scheduling algorithm based on the quasi ordering for related backup task on the scheduler, and we prove that this algorithm can find an optimum order for related backup task and the algorithm can be used to improve the performance of the whole system; finally, an experiment proves that our approach is successful. When there are about 200 storage backup tasks, the execution time can be saved about 20%, which means that the performance of the system is much improved by using our new scheduling algorithm.

关 键 词：虚拟存储 关联备份任务 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]