基于模糊层次分析的虚拟机热度综合评估方法  

Virtual Machine Hotspot Degree Comprehensive Evaluation Method Based on Fuzzy Analytic Hierarchy Process

作　　者：闫永明[1] 张斌[1] 郭军[1] 莫玉岩 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110819

出　　处：《东北大学学报（自然科学版）》2015年第2期182-187,共6页Journal of Northeastern University(Natural Science)

基　　金：中央高校基本科研业务费专项资金资助项目(N120804001;N120204003);国家自然科学基金资助项目(61300019)

摘　　要：虚拟机热点的判断是虚拟机热点消除过程中的关键环节.传统方法通常判断监测指标是否超过阈值,未考虑判断指标与服务之间的关系,影响判断的准确性.本文结合部署在虚拟机上的服务的可用性和质量因素建立了热度评估指标体系,提出了基于模糊层次分析的主观权重和基于离差最大化法的客观权重的确定算法,以及将主观权重和客观权重相结合的热度综合评估方法,并给出了基于热度的虚拟机冷热点判断规则.实验结果表明,利用热度进行热点判断的准确率高于利用传统设定阈值的方法,可以有效减少不适当的迁移,热点消除的代价较小.Evaluating the hotspot degree of VM （virtual machine） is the critical step of the VM hotspot eliminating process. Traditional approaches of evaluating hotspot degree often monitor whether one or more threshold values being exceeded or not. The correlation among evaluation metrics and the adaptation of different metrics were not considered in the existing approaches for various services, which influences evaluation accuracy. The hotspot degree evaluation system is created, combining VM service availability and service quality factors. The subjective weight algorithm based on fuzzy analytic hierarchy process is proposed, as well as the objective weight algorithm based on maximum deviation and hotspot degree comprehensive evaluation method based on these two weights. A rule of VM coldspot or hotspot evaluation based on hotspot degree is presented. The experiment result indicates that the accuracy of the proposed method is higher than that of traditional approaches in evaluating the hotspot degree, which can efficiently reduce improper migration, with lower prices.

关 键 词：云计算 热点消除 虚拟机热度 冷点 模糊层次分析法 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]