基于云科学工作流调度的代价与能效优化算法

为了降低云环境中科学工作流调度的执行代价与数据中心能耗,提出了一种基于能效感知的工作流调度代价最优化算法CWCO-EA。算法在满足截止时间约束下,以最小化工作流执行代价与降低能耗为目标,将工作流的任务调度划分为四步执行。首先,通过代价效用的概念设计虚拟机选择策略,实现了子makespan约束下的任务与最优虚拟机间的映射；其次,通过串行与并行任务合并策略,同步降低了工作流的执行代价与能耗；然后,通过空闲虚拟机重用机制,改善了租用虚拟机的利用率,进一步提高了能效；最后,通过任务松驰策略实现了租用虚拟机的能力回收,节省了能耗。通过四种科学工作流的仿真实验,结果表明,CWCO-EA算法比较同类型算法,在满足截止时间的同时,可以同步降低工作流的执行代价与执行能耗。

Cost and Energy-efficiency Optimization Algorithm Based on Cloud Scientific Workflow Scheduling

In order to reduce the execution cost and energy consumption of data center in scientific workflow scheduling under cloud environment, a workflow scheduling cost optimization algorithm based on energy-efficiency aware CWCO-EA is presented in this paper. Under satisfying deadline constraint, our algorithm defines minimizing the workflow execution cost and reducing energy consumption as the objectives, and divides the task scheduling of workflow into four steps. First, we use the virtual machine (VM) selection strategy with applies the concept of cost utility to map tasks to their optimal VM types by the sub-makespan constraint. Second, the sequence tasks and parallel tasks merging strategies are employed to reduce execution cost and energy consumption of workflow. Then, we use the idle VM reuse mechanism to improve the utility of leased VMs and further improve the energy-efficiency. Finally, the scheme of slack time reclamation is utilized to save energy of leased VM. Through the simulation experiments of four real-world scientific workflow, the results show that, compared with the same types of related well-known algorithms, under meeting the deadline constraint, CWCO-EA can synchronously reduce the workflow execution cost and energy consumption.