面向云计算服务的能耗采集与预测方法

发布时间：2018-04-11 05:35

本文选题：云计算 + 能耗采集　；参考：《华东师范大学》2016年硕士论文

【摘要】：云计算作为一种新的计算资源提供模式,在商业系统特别是面向大量并发用户以及复杂计算任务的商业业务服务系统中具有广泛的应用前景。随着云计算资源规模的不断扩增,云数据中心的能耗花费在数据中心运营成本中所占的比例日益增加,云数据中心能耗管理问题变得越来越突出,也得到了行业内越来越多的关注。在云环境下,依托于物理机之上的虚拟机是云服务程序的运行的基础环境。然而,在不同的资源管理策略下,服务程序与虚拟机的运行能耗有着很大的差异。目前,云环境下能耗研究对象主要是虚拟机,而忽略了运行在其之上的服务程序。为了对运行在云环境下的服务程序进行能耗研究,本文提出了一个有效的云环境下面向云计算服务的能耗测试框架。本文提出的云环境下能耗测试框架实现了对云环境基础平台、虚拟机以及运行在虚拟机之上的服务程序进行能耗监控与测量。通过有效的测试和分析,对整个运行环境进行层层分离以去除云环境中物理机以及虚拟机的基准能耗,从而获取执行服务程序本身所需的准确能耗。通过此测试框架,本文对不同资源管理策略下的能耗进行了测试,对结果的差异性进行分析研究,并在此基础上进行准确的能耗建模和预测。利用一个基于Open Stack的云环境,我们通过一个详细的案例进行能耗数据的采集,并对采集得到的能耗数据进行建模,利用时序模型进一步实现了对已知服务程序和未知服务程序在云环境下运行所需能耗的预测。实验结果证明本文提出的能耗测试框架和能耗预测方法具有较好的通用性和准确性。
[Abstract]:Cloud computing, as a new computing resource providing mode, has a wide application prospect in commercial systems, especially in commercial business service systems, which are oriented to a large number of concurrent users and complex computing tasks.With the expansion of cloud computing resource scale, the energy consumption of cloud data center is increasing in the operation cost of data center, and the problem of energy consumption management of cloud data center becomes more and more prominent.Also got more and more attention in the industry.In the cloud environment, the virtual machine relying on the physical machine is the basic environment for the cloud service program to run.However, under different resource management strategies, the running energy consumption of service program and virtual machine is very different.At present, the research object of energy consumption in cloud environment is virtual machine, but ignore the service program running on it.In order to study the energy consumption of service programs running in the cloud environment, this paper proposes an effective energy consumption testing framework for cloud computing services.The energy consumption testing framework proposed in this paper can monitor and measure the energy consumption of the basic platform, virtual machine and service program running on the cloud environment.Through effective test and analysis, the whole running environment is separated layer by layer to remove the benchmark energy consumption of the physical machine and virtual machine in the cloud environment, so as to obtain the accurate energy consumption needed to execute the service program itself.Through this testing framework, the energy consumption under different resource management strategies is tested, the difference of results is analyzed and studied, and on the basis of this, accurate energy consumption modeling and prediction are carried out.Using a cloud environment based on Open Stack, we collect energy consumption data through a detailed case, and model the energy consumption data.The time series model is used to predict the energy consumption of the known and unknown service programs in the cloud environment.The experimental results show that the proposed energy test framework and the energy consumption prediction method are of good generality and accuracy.
【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.09

【相似文献】