Abstract: The energy consumption of servers produced in tasks executing is an important part of system dynamic energy in cloud computing system. In order to lower the total task execution energy consumption of system, this paper presents a task scheduling method based on energy optimization with the earliest task completion time, builds the dynamic power computing model of server, energy model based on dynamic power of servers, and the energy optimization model of cloud computing system. The proposed method uses different scheduling algorithms to get less task execution energy according to the deadline of task and the execution energy on each server. Experimental results show that this scheduling policy can better meet the deadline of task, and reduce the total energy of cloud computing system considerably.

Key words: cloud computing, energy optimization, task scheduling, task deadline

摘要： 服务器执行任务产生的能耗是云计算系统动态能耗的重要组成部分。为降低云计算系统任务执行的总能耗，提出了一种基于能耗优化的最早完成时间任务调度方法，建立了服务器动态功率计算模型，基于动态功率的服务器执行能耗模型，以及云计算系统的能耗优化模型。调度策略根据任务的截止时间要求和在不同服务器上的执行能耗，选择不同的调度算法，以获得最小任务执行总能耗。实验结果证明，提出的任务调度方法，能够较好地满足任务截止时间的要求，降低云计算系统任务执行的总能耗。

关键词: 云计算, 能耗优化, 任务调度, 任务截止时间