摘要
传统虚拟机部署侧重降低主机能耗,忽略了全局能效。针对这一问题,提出一种自适应多重阈值的虚拟机部署与迁移优化算法。基于主机CPU利用率的历史数据集,设计两种基于K-均值聚簇的自适应多重阈值决策方法,依据多重阈值对主机进行分类;为对重载主机进行虚拟机迁移,设计3种虚拟机迁移选择方法,以能效最高的方式对迁移虚拟机进行重新部署。通过实际负载数据对算法进行仿真测试,测试结果表明,该算法可以有效降低能耗,SLA违例也较低,具有更高的能效。
Traditional virtual machines placement methods focus on reducing energy consumption on hosts without considering the overall energy-efficiency improvement.Aiming at this problem,a virtual machine placement and migration optimization algorithm based on adaptive multi-threshold was presented.Based on the historical data set of CPU utilization on hosts,two adaptive multi-thresholds decision methods based on K-means clustering were designed.According to the multi-threshold,all hosts were divided.For migrating some virtual machines from heavy hosts,three virtual machines migration selection methods were designed and the migrated virtual machines with highest energy-efficiency idea were re-allocated.Some extensive comparison experiments were performed using real-world workload.The results show that,the proposed algorithm can reduce the energy cons- umption while maintaining low SLA violation,which has higher energy-efficiency.
作者
张磊
王莉
ZHANG Lei;WANG Li(Department of Software and Communication,Tianjin Sino-German University of Applied Sciences,Tianjin 300350,China)
出处
《计算机工程与设计》
北大核心
2019年第8期2216-2223,共8页
Computer Engineering and Design
基金
国家重点研发计划基金项目(2017YFC0804301)
天津市教委科研计划基金项目(2017KJ040)
天津市企业科技特派员基金项目(18JCTPJC49700、18JTPC50000)
关键词
云数据中心
虚拟机部署
虚拟机迁移
能效优化
服务等级协议
cloud data center
virtual machine placement
virtual machine migration
energy-efficiency optimization
SLA
