网格计算中基于任务优先级的负载均衡算法  

Based on parity and load-balance task scheduling algorithm in computing gird

作　　者：魏妮妮 宋翌[1] 艾学轶[1] 

机构地区：[1]武汉生物工程学院计算机与信息工程系,武汉430415

出　　处：《华中师范大学学报（自然科学版）》2013年第2期173-177,共5页Journal of Central China Normal University：Natural Sciences

基　　金：湖北省自然科学基金项目(2011CD297);新世纪优秀人才支持计划项目(NCET-08-0806);武汉市教育局科研项目(2009k100)

摘　　要：任务调度是网格计算研究的核心问题之一,任务调度策略的好坏将直接影响网格系统的性能.在对已有启发式任务调度算法进行分析比较基础上,针对于网格计算中的高吞吐率应用的任务调度问题,提出了一种综合考虑资源节点性能度量和任务优先级的负载均衡的启发式任务调度算法.进行了仿真实验,在任务完成时间和系统吞吐率二个方面与传统的Min-min算法、Max-min算法和Sufferage算法进行比较.仿真结果表明该算法不但可以减小任务总的完成时间,而且可以提高系统的吞吐率,更适用于高吞吐率应用任务调度问题的解决.Task scheduling is one of the core issues of grid computing, scheduling strate- gy will directly affect the performance of grid system. Based on analysis and comparison to the existing heuristic scheduling algorithm, for the task scheduling problem of the computing grid which with the high-throughput application, proposed a algorithm based on comprehensive consideration of resources performance metrics and task node needs, the algorithm based on task priority and load-balancing task scheduling algorithm. With a simulation experiment, compare with the traditional three aspects of Min-min algorithm Max-rain algorithm and Sufferage comparison algorithm in the task completion time and system resource utilization. The simulation results show that this algorithm not only makes the overall task completion time are reduced, but also improve the utilization of system resources, so it proves that the algorithm is effective, and more suitable for applications in high-throughput scheduling.

关 键 词：网格计算 任务调度 性能预测 优先级 负载均衡 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]