云计算系统能耗优化方法的研究

发布时间：2018-11-22 14:16

【摘要】：高能耗是云计算系统的一个主要问题,且随着云计算规模的日益扩大,其能量消耗越来越严重。为了减少云计算系统的能量消耗,针对云计算系统中的空闲能耗和奢侈能耗,进行了能耗优化方法的研究。首先,对传统能耗优化技术进行了总结,其中重点分析了关闭/休眠技术和资源调度算法在云计算系统能耗优化领域的研究现状。其次,综合考虑了云计算系统中设备的不同状态及任务的不同类型,建立了云计算系统能耗模型MSEM (Multi-State Energy Model),并通过实验对模型的有效性进行了验证。针对空闲能耗,在指数平均算法的基础上进行改进,提出了一种自适应的空闲时间预测策略。该策略引入了自适应权值调节因子,动态调节历史空闲时间对预测空闲时间的影响,同时引入分段滑动窗口的思想,将滑动窗口内的空闲时间分为长、中、短三类,减少了偶然波动对预测结果的不良影响。实验结果表明提出的策略预测准确率较高,系统响应时间少,节能效果好。针对奢侈能耗,以Min-Min算法作为基础,在满足QOS的前提下,考虑能耗及负载平衡因素,对算法进行改进,提出了一种资源调度方法Energy-Aware Min-Min。并在云计算仿真平台CloudSim上进行仿真实验,实验结果表明资源调度算法Energy-Aware Min-Min可以达到能耗优化的目的。
[Abstract]:High energy consumption is a major problem in cloud computing systems, and with the increasing scale of cloud computing, its energy consumption is becoming more and more serious. In order to reduce the energy consumption of cloud computing system, the energy consumption optimization method is studied for idle energy consumption and luxury energy consumption in cloud computing system. Firstly, the traditional energy consumption optimization technology is summarized, and the research status of shutdown / sleep technology and resource scheduling algorithm in the field of energy consumption optimization in cloud computing system is analyzed. Secondly, considering the different states of devices and different types of tasks in cloud computing system, the energy consumption model MSEM (Multi-State Energy Model),) of cloud computing system is established, and the validity of the model is verified by experiments. Based on the exponential average algorithm, an adaptive idle time prediction strategy is proposed for idle energy consumption. The strategy introduces adaptive weight adjustment factor, dynamically adjusts the influence of historical idle time on prediction of idle time, and introduces the idea of piecewise sliding window. The idle time in sliding window is divided into three categories: long, medium and short. The adverse effects of accidental fluctuations on the forecast results are reduced. The experimental results show that the proposed strategy has higher prediction accuracy, less system response time and better energy saving effect. Aiming at luxury energy consumption, on the basis of Min-Min algorithm and considering the factors of energy consumption and load balance, this paper proposes a resource scheduling method, Energy-Aware Min-Min., on the premise of satisfying QOS and considering the factors of energy consumption and load balance. The simulation results on the cloud computing simulation platform CloudSim show that the resource scheduling algorithm Energy-Aware Min-Min can achieve the purpose of energy consumption optimization.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】