云计算中资源延迟感知的实时任务调度方法

发布时间：2018-04-28 06:02

本文选题：云计算 + 虚拟化　；参考：《计算机研究与发展》2017年02期

【摘要】：绿色云计算已经成为一个研究焦点,动态整合虚拟机和关闭空闲主机是极具潜力的途径可降低云计算数据中心的能耗.当云平台的负载迅速增加时,系统需要启动更多的主机和创建更多的虚拟机来扩展可用资源.然而,启动主机和创建虚拟机需要一定的时间开销,使得紧急任务难以及时开始,从而延误了截止期.为了解决以上问题,首先提出具有机器启动时间感知的虚拟机扩展策略,以缓解机器启动时间冲击实时任务的时效性要求.基于该策略,设计算法STARS来调度实时任务和资源,以在保障任务时效性与节能2方面进行权横.最后,使用Google的负载数据进行模拟实验,比较算法STARS与其他2个算法的性能.实验结果表明,在保障任务时效性、节能和资源利用率方面,算法STARS优于对比算法.
[Abstract]:Green cloud computing has become a research focus. Dynamic integration of virtual machines and shutdown of idle hosts are potential ways to reduce energy consumption in cloud computing data centers. When the load on the cloud platform increases rapidly, the system needs to start more hosts and create more virtual machines to extend the available resources. However, starting the host and creating a virtual machine requires a certain amount of time overhead, which makes it difficult to start the urgent task in time, thus delaying the deadline. In order to solve the above problems, a strategy of virtual machine extension with machine start time awareness is proposed to alleviate the timeliness requirement of machine startup time impact on real time tasks. Based on this strategy, the algorithm STARS is designed to schedule real-time tasks and resources to ensure task timeliness and energy saving. Finally, the load data of Google are used to simulate and compare the performance of the algorithm STARS and the other two algorithms. The experimental results show that the algorithm STARS is superior to the contrast algorithm in ensuring task timeliness, energy saving and resource utilization.
【作者单位】：国防科学技术大学信息系统工程重点实验室;
【基金】：国家自然科学基金项目(61572511,71271213) 国防科学技术大学科研计划项目(ZK16-03-09)~~
【分类号】：TP3

【相似文献】