摘要
任务调度技术是并行分布式系统中的关键技术之一,对系统的性能起着重要作用,但通常情况下大型系统的任务调度问题属于NP问题。而现代启发式生物进化算法是找出很多NP问题近似解的有效方法。本文将粒子群算法应用于基于可用性的网格系统调度中,提出了一种调度算法,对算法的性能进行了理论分析和模拟实验。结果表明:和最近文献中的基于可用性的调度算法SSAC相比,所提出的新算法在保证系统资源具有同样的可用性条件下,能够产生更好的调度长度。
Task scheduling techniques, playing an important role to system performance, are one of the key technology in parallel and distributed systems. In general, the task scheduling problem in large- scale systems is a NP problem. The modern heuristic biological evolution algorithm is an effective meth- od to find an approximate solution for NP problems. In this paper, a scheduling algorithm which applys particle swarm algorithms into availability grid dispatch systems is proposed. The performance of the al- gorithm is analysed in theory. The simulation results show that compared with the SSAC algorithm which is recently proposed, the proposed particle swarm optimization task scheduling algorithm can gen- erate shorter makespan while conserving the same availability level.
出处
《计算机工程与科学》
CSCD
北大核心
2012年第6期44-49,共6页
Computer Engineering & Science
基金
国家自然科学基金资助项目(90715029
61070057)
教育部新世纪优秀人才支持计划资助(NCET-08-0177)
关键词
并行与分布式处理
异构系统
任务调度
可用性
粒子群算法
parallel and distributed computing
heterogeneous system
task scheduling
availability
particle swarm optimization algorithm
