基于数据中心的两阶段虚拟机能效优化部署算法

针对数据中心在虚拟机动态部署过程中的高能耗问题,提出了面向数据中心的两阶段虚拟机能效优化部署算法--DVMP_VMMA。第一阶段为初始部署,提出了动态虚拟机部署(DVMP)算法限定主机最优部署数量,降低了闲置能耗;同时,为了应对负载的动态变化,第二阶段提出迁移约束的虚拟机迁移算法(VMMA)对初始部署方案作进一步优化,这样不仅得到的系统能耗更低,而且还能保证应用服务质量。与满载算法(FL)、基于固定门限值的部署算法(FT),绝对中位差部署算法(MAD)、四分位差部署算法(QD)、迁移周期最优算法(MTM)、最小占用率迁移算法(MIU)进行的比较实验结果表明:DVMP_VMMA不仅考虑了系统能耗优化,使运行时资源利用率更高;而且还可以避免VM频繁迁移完成对性能的提升,其在优化数据中心能耗、SLA违例、VM迁移量的控制及性能损失等指标上均有较好效果,其综合性能优于对比算法。

Two-phase placement algorithm with energy efficiency optimization for virtual machins bsed on data center

In view of high energy consumption of dynamic virtual machine placement in data centers, a two-phase placement algorithm with energy efficiency optimization for virtual machines, named DVMP_VMMA, was proposed based on data center. The first phase was the initial placement, a Dynamic Virtual Machine Placement (DVMP) algorithm was presented to limit the optimal number of deployed hosts, which reduced the idle energy consumption. Meanwhile, for responding to the dynamic changes of loads, Virtual Machine Migration Algorithm (VMMA) was put forward to further optimize the initial placement with the migration constraints through the dynamic VM (Virtual Machine) migration in second phase, which not only got lower energy consumption of the system, but also ensured Quality of Service (QoS) of applications. Comparison experiments with FL (Full Load), FT (Fixed Threshold), MAD (Median Absolute Deviation), QD (Quartile Deviation), MTM (Migration Time Minimum) and MIU (Minimum Utilization) were given. The experimental results show that DVMP_VMMA not only considers the energy consumption optimization to increase the resource utilization, but also avoids frequent migration of VMs to improve the performance, it gets good effect in optimization of data center energy consumption, SLA (Service Level Agreement) violation, VM migration quantity and loss of performance, its comprehensive performance is better than compared algorithms.