摘要
云计算环境下,在满足用户服务级目标约束下,如何有效地进行资源分配调度,降低能耗,已成为不容忽略的关键问题.针对目前云计算系统服务资源分配调度问题在能耗方面的研究不足,提出一种能耗优化的资源分配调度体系架构,并基于此架构设计了一个满足实时用户SLA的能耗优化模型.该优化模型从系统级和部件级两个层次进行能耗优化.在系统级上,提出一种基于分组遗传算法最大限度降低系统空闲能耗的虚拟机部署算法,该算法将虚拟机和服务器之间的映射抽象为有约束的多维可变装箱问题;同时,在部件级上采用动态电压功率调整技术降低执行能耗,从而达到在满足用户需求的前提下,最大限度降低系统总能耗.仿真实验结果表明,该算法与同类算法相比,在相同条件下可有效降低云计算系统的能耗开销.
In cloud computing, how to allocate and schedule resources effectively and how to reduce energy consumption under the restraint of satisfying the customer service level have become key issues that cannot be ignored. Since there is a lack of studies on energy consumption by resources allocation and scheduling, we propose a new resources-allocation and scheduling architecture for energy consumption optimization. Based on this architecture, a new energy consumption optimization model is designed to meet the real-time service level agreement （SLA）. The proposed model optimize energy consumnption both on the system level and the component level On the former level, a new virtual machine deployment algorithm based on grouping genetic algorithm is proposed to minimize systems＇ idle energy consumption, which abstract the mapping between virtual machines and Servers into a multidimensional variable packing problem. On the latter level, dynamic voltage power adjustment technology is used to reduce execution energy consumption. Therefore, energy consumption can be minimized on the both levels with premise of meeting users＇ requirements. Experimental results show that compared with other algorithms, the proposed one can greatly reduce the total energy consumption of cloud computing systems under the same conditions.
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2016年第3期768-778,共11页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(U1204618
U1404622)
中国博士后科学基金(2012M512008)
河南省科技发展计划基金(152102310381)~~
关键词
云计算
实时任务
能耗优化
部署算法
cloud computing
real-time task
energy consumption optimization
deployment algorithm
