摘要
对网格计算环境中的任务调度模块做了分析,介绍了两种最经典的网格任务调度算法:Min-Min与Max-Min。通过对两者的分析,得知Min-Min算法的优点在于其执行的高效性,但它存在资源负载不平衡的问题。而Max-Min则正好相反。经过对比,提出了一种平衡两者优缺点的A-MM算法。使用GridSim网格环境仿真包建立A-MM调度算法的模拟平台,对GridSim的内部机制及用法给予简单介绍,并通过模拟实验对A-MM算法进行性能评测。评测结果表明:A-MM算法的完成时间接近Min-Min,而负载平衡性接近Max-Min,平衡了Min-Min与Max-Min的优缺点。
Analyzes the task scheduling module in grid computing environment, and introduces two of the most classic grid task scheduling algorithm： Min-Min and Max-Min. Based on the analysis of the two, that the Min-Min algorithm advantage lies in its implementation of efficiency, but it exists a problem that resources load imbalance. Max-Min algorithm is for the very opposite. After contrast,introduces a new algorithm named A-MM which could balance the advantages and disadvantages of the two. Uses GridSim toolkit to establish simulation platform of A-MM,and introduces internal mechanisms and usage of GridSim, and uses simulation experiments to evaluate the performance of A-MM. Evaluation results show that：the completion time of A-MM algorithm close to Min-Min,and the load balancing of A-MM algorithm close to Max-Min. A-MM algorithm balances Min-Min and Max-Min＇s strengths and weaknesses.
出处
《计算机技术与发展》
2008年第10期96-98,102,共4页
Computer Technology and Development
基金
国家自然科学基金(60563002)
新疆自治区高校科研重点项目(XJEDU2004I03)
