面向云计算数据中心的能耗建模方法

发布时间：2018-05-03 19:05

本文选题：云计算 + 数据中心　；参考：《软件学报》2014年07期

【摘要】：云计算对计算能力的需求,促进了大规模数据中心的飞速发展.与此同时,云计算数据中心产生了巨大的能耗.由于云计算的弹性服务和可扩展性等特性,云计算数据中心的硬件规模近年来极度膨胀,这使得过去分散的能耗问题变成了集中的能耗问题.因此,深入研究云计算数据中心的节能问题具有重要意义.为此,针对云计算数据中心的能耗问题,提出了一种精确度高的能耗模型来预测云计算数据中心单台服务器的能耗状况.精确的能量模型是很多能耗感知资源调度方法的研究基础,在大多数现有的云计算能耗研究中,多采用线性模型来描述能耗和资源利用率之间的关系.然而随着云计算数据中心服务器体系结构的变化,能耗和资源使用率的关系已经难以用简单的线性函数来描述.因此,从处理器性能计数器和系统使用情况入手,结合多元线性回归和非线性回归的数学方法,分析总结了不同参数和方法对服务器能耗建模的影响,提出了适合云计算数据中心基础架构的服务器能耗模型.实验结果表明,该能耗模型在只监控系统使用率的情况下,在系统稳定后,能耗预测精度可达到95%以上.
[Abstract]:The demand for computing power in cloud computing promotes the rapid development of large-scale data centers. At the same time, cloud computing data centers have generated huge energy consumption. Due to the flexibility and scalability of cloud computing, the hardware scale of cloud computing data center has expanded extremely in recent years, which makes the decentralized energy consumption problem become centralized energy consumption problem in the past. Therefore, it is of great significance to study the energy saving of cloud computing data center. Aiming at the energy consumption of cloud computing data center, a high precision energy consumption model is proposed to predict the energy consumption of a single server in cloud computing data center. Accurate energy model is the basis of many energy consumption aware resource scheduling methods. In most of the existing cloud computing energy consumption research, linear model is used to describe the relationship between energy consumption and resource utilization. However, with the change of cloud computing data center server architecture, the relationship between energy consumption and resource utilization is difficult to describe with a simple linear function. Therefore, starting with processor performance counter and system usage, combining the mathematical methods of multiple linear regression and nonlinear regression, the influence of different parameters and methods on the modeling of server energy consumption is analyzed and summarized. A model of server energy consumption suitable for cloud computing data center infrastructure is proposed. The experimental results show that the accuracy of energy consumption prediction can reach more than 95% when the system is stable and only the utilization rate of the system is monitored.
【作者单位】：电子科技大学计算机科学与工程学院;北京航空航天大学计算机学院;
【基金】：国家高技术研究发展计划(863)(2013AA01A210)
【分类号】：TP308

【参考文献】