摘要
Hadoop作为世界领先的大数据平台,其性能更多地依赖于Map Reduce任务调度机制。通过对Map Reduce任务调度机制中推测算法的研究,提出一种高效、准确和基于优先级的改进Hadoop调度算法。通过测试发现,改进后的Hadoop调度算法在异构环境下能够对落后任务判定准确,更好地维持系统的负载平衡,减少系统对任务的响应时间,增加对高优先级任务的响应速度,提高Map Reduce任务调度算法的性能。
As the world＇s leading data platform, Hadoop＇s performance deeply depends on the Map Reduce scheduling mechanism.In this paper, through a speculative algorithm research on Map Reduce scheduling mechanism, an efficient, accurate and priority-based advanced Hadoop scheduling mechanism algorithm is proposed. This algorithm can make exactly judgment to backward task in heterogeneous environment by testing. It maintains the balance of the system load better and reduces the system response time to the tasks. The algorithm also improves the response speed to the tasks of high priority and the performance of Map Reduce scheduling mechanism.
出处
《微型电脑应用》
2015年第6期55-58,共4页
Microcomputer Applications
关键词
异构环境
推测算法
负载均衡
优先级
Heterogeneous Environment
Speculative Algorithm
Load Balancing
Priority
