摘要
在云数据中心,虚拟机整合(VMC)是绿色计算和最小化集群功耗问题的关键技术。大多数研究采用基于资源使用率来寻求最合理的虚拟机(VM)整合方式,但在整合过程中并未考虑物理服务器(PM)的可用性及同位VM的稳定性问题。本文提出了基于互斥条件限制的VMC策略,该策略同时考虑PM的可用性和VM之间的互斥性两方面因素;给出了集群服务器统一资源预留计算方法,用于保证PM可用性的问题,并给出了基于VM历史运行数据的相似度判定方法,用于解决互斥条件的判定问题。将该策略应用于运行在OpenStack平台中的VM数据,实验结果表明:该策略可以有效地保证PM的可用性,避免同类型VM被整合在一起,减少同位VM的性能损失和确保服务质量。
In the cloud data center,virtual machine consolidation(VMC)is one of the key technologies for green computing and minimizing cluster power consumption.In the relevant research works,most seek the most reasonable consolidation schemes based on resource utilization and don't consider the stability and availability of PM and co-located virtual machines(VM).In this paper,by analyzing their mutual exclusion conditions,a virtual machine consolidation strategy is proposed,in which both the availability of PM and the mutual exclusivity among VM are considered.Moreover,the cluster server uniform resources reservation approach and the similarity calculation method based on the historical running data of VM are also given.Finally,this strategy is evaluated by collecting and analyzing the running data of VM deployed on the open stack platform.Experiment results show that the proposed method can effectively guarantee the PM availability and avoid the situation that the same type of VM is consolidated in one PM.Hence,this algorithm can reduce the performance degradation of co-located VM and improve the quality of service(Qos).
出处
《华东理工大学学报（自然科学版）》
CAS
CSCD
北大核心
2017年第1期119-128,共10页
Journal of East China University of Science and Technology
基金
国家自然科学基金(61173048
61300041
61472139)
高等学校博士学科点专项科研基金博导类资助课题(20130074110015)
中央高校基本科研业务费专项基金(WH1314038
WH1514331)
关键词
互斥条件
虚拟机整合
相关性
云数据中心
mutual exclusion conditions
virtual machine consolidation
correlation
cloud data center
