虚拟化云中随机任务与资源调度方法

任务和资源调度方法是云系统的关键技术之一。但是,现有的研究往往忽略实时任务的高动态性和任务执行时间的随机性,使得调度方案的实际性能与期望性能相差甚远。针对以上问题,本文设计一个随机性感知的调度框架;提出一个启发式调度算法集成前摄性和反应式策略(proactive and reactive strategy, PRS)来对任务进行调度,以提高云系统保障实时任务时效性的能力;并提出3个计算资源伸缩策略来动态调整计算资源,以减少能量消耗。最后,通过实验将算法PRS的性能与其他4个算法进行比较。实验结果表明,在任务完成率和能耗方面,算法PRS的性能比已有算法提高13.85%和17.23%。

Scheduling for stochastic tasks and resources in virtualized clouds

Task and resource scheduling is one of the key technologies for the cloud system. However, the existing research tends to ignore the dynamic nature of real-time tasks and the randomness of task execution time, which makes the pre-generated schedule may not being effective in real execution. To address this issue, a randomness aware scheduling architecture is designed; a heuristic scheduling algorithm, integrating proactive and reactive strategy （PRS）, is proposed to schedule tasks dynamically, which improves the ability of the cloud system to guarantee the timeliness of real-time tasks; three strategies are proposed to scale up/down computing resources according to the system load to reduce the energy consumption. Finally, the performance of the algorithm PRS is compared with the other four algorithms. The experimental results show that compared with the existing algorithms the performance of the algorithm PRS is improved by 13.85% and 17.23% in terms of guarantee ratio and energy consumption.