Abstract: By studying the colony algorithm, analyzing the deficiency of existing Hadoop scheduler, a resource-aware scheduler based on ant colony algorithm is proposed. It  can be more comprehensive and more scientific to assign tasks and improve the operating performance of the overall architecture. The experiment shows that, the resource-aware scheduler has not improved the performance markedly in isomorphic situation, however, it can significantly increase performance in isomerism situation. This resource-aware scheduler can improve the calculation speed of the system and reduce the computing time.

Key words: Hadoop, resource-aware scheduler, ant colony algorithm

摘要： 通过研究蚁群算法，针对现有Hadoop调度器的不足，提出一个基于蚁群算法的Hadoop资源感知调度器及其具体实现方案。从而使Hadoop作业调度器可以更有效地对任务进行分配，提高整体架构的作业性能。通过实验证明，利用蚁群算法实现的资源感知调度器在同构环境中虽没有明显改善系统计算速度，但是在异构环境中可以很好提高系统处理任务的性能，降低了运算时间。

关键词: Hadoop, 资源感知调度器, 蚁群算法