云平台上基于关键路径截取的有向无环图应用调度算法

发布时间：2018-02-25 10:17

本文关键词： 云计算平台关键路径虚拟机有向无环图资源配置　出处：《国防科技大学学报》2017年03期 　论文类型：期刊论文

【摘要】：针对云平台上有向无环图科学应用执行容易产生虚拟机资源过剩、资源使用率低及费用虚高的问题,给出一种基于关键路径截取的有向无环图应用调度算法。该算法采取关键路径截取技术,循环找出最晚完成的未分配任务,从该任务出发,在所有未分配任务构成的图中找出最大连通子图,并计算该子图的关键路径,然后将关键路径上的任务集调度到性能匹配的虚拟机上执行;同时通过任务回填技术充分利用虚拟机的空闲时间槽,提高资源使用率。实验结果表明,在云计算平台上,该算法不仅能够在截止时间内完成有向无环图科学应用,而且可以提高资源使用率,有效减少完成该应用所需整体费用。
[Abstract]:In view of the problem that the scientific application of directed acyclic graph on the cloud platform can easily lead to the surplus of virtual machine resources, the low utilization rate of resources and the high cost of virtual machine, A directed acyclic graph application scheduling algorithm based on critical path interception is presented. The maximum connected subgraph is found in all unassigned task graphs, and the critical path of the subgraph is calculated, then the task set on the critical path is scheduled to execute on the performance matching virtual machine. At the same time, the task backfilling technology is used to make full use of the idle time slot of the virtual machine to improve the resource utilization. The experimental results show that the algorithm can not only accomplish the scientific application of directed acyclic graph within the deadline, but also improve the resource utilization. Moreover, it can increase the utilization rate of resources and reduce the overall cost of completing the application.
【作者单位】：国防科技大学计算机学院;国防科技大学海洋科学与工程研究院;
【基金】：国家自然科学基金资助项目(61572510) 国家公益行业专项计划资助项目(GYHY201306003)
【分类号】：TP301.6;TP393.09

【相似文献】