基于K最近邻回归预测的高能效虚拟机合并  

High energy-efficient virtual machine consolidation based on K-nearest neighbor regression forecasting

作　　者：王诺 李艳[1] WANG Nuo;LI Yan(Center of Information Management,Hebei Institute of Communications,Shijiazhuang 051430,China)

机构地区：[1]河北传媒学院信息管理中心,河北石家庄051430

出　　处：《计算机工程与设计》2021年第5期1235-1243,共9页Computer Engineering and Design

基　　金：国家重点研发计划基金项目(2017YFC0804301)。

摘　　要：虚拟机合并和迁移仅考虑当前负载会导致过多非必要迁移,为此,提出基于资源利用预测的虚拟机合并算法UP-BFD。通过K最近邻回归方法同时对主机和虚拟机的负载进行预测,在虚拟机迁移源主机和目标主机的选择上,同步考虑当前超载和预测超载问题,较好避免无用虚拟机迁移。通过随机负载和现实负载进行仿真测试,测试结果表明,UP-BFD算法可以降低主机总体能耗,同步减少SLA违例和虚拟机迁移量。Virtual machine consolidation and migration considering only current workload leads to excessive unnecessary migrations.To solve this problem,a virtual machine consolidation algorithm based on resource utilization forecasting was proposed,called UP-BFD.The load of the host and virtual machine was forecasted simultaneously through K-nearest neighbor regression,and the current overload and predicted overload were synchronously considered in the selection of the virtual machine migration source host and target host,thus avoiding the useless virtual machine migration.The algorithm was simulated by the random load and the real workload.The results show that UP-BFD can not only reduce the overall energy consumption of the host,but synchronously minimize SLA violations and virtual machine migration.

关 键 词：云计算 预测模型 虚拟机迁移 K最近邻回归 能效 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]