北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2016, Vol. 39 ›› Issue (1): 107-111.doi: 10.13190/j.jbupt.2016.01.020

• 研究报告 • 上一篇    下一篇

云计算环境下的非线性能耗建模方法

肖丁, 贾亚璞, 石川   

  1. 北京邮电大学通信软件工程中心, 北京 100876
  • 收稿日期:2015-09-16 出版日期:2016-02-28 发布日期:2016-01-29
  • 作者简介:肖丁(1966-),男,硕士,讲师,E-mail:dxiao@bupt.edu.cn.
  • 基金资助:

    国家重点基础研究发展计划(973计划)项目(2013CB329606)

Non-liner Energy Consumption Model for Cloud Computing

XIAO Ding, JIA Ya-pu, SHI Chuan   

  1. Telecommunication Software Engineering Group, Beijing University of Posts and Telecommunications, Beijing 100876, China
  • Received:2015-09-16 Online:2016-02-28 Published:2016-01-29

摘要:

针对数据中心的计算节点建立精密的能耗模型来预测能耗状况很有必要.为了解决线性模型准确性不足的问题,提出了一种基于分段式CPU利用率的能耗模型,并且加入了内存和硬盘因素.在实验中发现了其他部件的耗能与CPU的内在联系,给出了潜在影响公式.随后,在单机环境和并行环境下验证了模型的准确性.

关键词: 能耗模型, 云计算, 非线性模型, 回归分析

Abstract:

With development of cloud computing and big data, the energy consumption of data center becomes more and more serious. It is necessary to establish a precise model of energy consumption for the computing node of the data center. In the most previous studies, linear system utilization is taken to establish the models of energy consumption. In order to solve the accuracy insufficient of linear model, a model of energy consumption based on the segmented CPU utilization was proposed when considering the memory and the hard drive at the same time. In the experiment, the inner relationship between the energy consumption and the CPU was found, and a formula of the potential influence of the components was proposed, and the accuracy of the model is verified in both single environment and parallel environment.

Key words: energy model, cloud computing, non-linear model, regression analysis

中图分类号: