异构云环境下能效优化的任务调度算法

摘要

伴随着全球数据中心和IT行业系统的电能消耗的不断攀升,能效优化的解决已经成为了当前在云数据中心系统中亟待解决的首要问题.就目前来看已提出的成熟的能效优化调度方法大部分针对的是同构服务器的方向,并不能较好地降低异构云环境下(如云数据中心)服务器的能耗.为解决此问题,论文对异构物理服务器引入了异构能效约束机制,经过深入研究建立了面向云数据中心的异构服务器能效优化资源的分配模型,并且提出了资源等价优化的求解方法,通过对资源分配搜索时对等价资源进行剪枝处理,使资源分配模型的求解空间缩小和模型的求解速度加快;深入探讨了面向划分服务器集群的可扩展分布式调度的方法以降低大规模云资源调度的复杂程度,实现了面向云数据中心的可扩展的高能效的资源分配模式的构建;以在前期云任务调度的研究基础之上,深入展开研究集成能效的优化和性能优化的云任务调度模型与算法,建立了面向不同需求的任务调度模式,使异构云环境下任务调度的能效比有了显著提高.%With the increase of power consumption in the global data center and IT industry,the optimization of energy effi-ciency has become the most important problem to be solved in the cloud data center system.For the time being,most of the mature energy efficient optimization scheduling methods are aimed at the direction of isomorphic server,and can not reduce the energy con-sumption of heterogeneous cloud environment(cloud data center)server better.In order to solve this problem,this paper introduces the heterogeneous energy restriction mechanism for heterogeneous physical server,after in-depth research establishes energy effi-ciency optimization of resources allocation in heterogeneous server model for cloud data center,and puts forward the solving method of equivalent resource optimization,resource allocation based on the search for equivalent resources for pruning treatment,the speed of solving the solution space the resource allocation model and the reduction of model complexity increases,discusses the method of scalable distributed scheduling for division of server cluster to reduce large-scale cloud resource scheduling,and build scalable energy-efficient resource allocation model for cloud data center,to research foundation in the early stage of cloud task scheduling.Of integrated energy efficiency optimization and performance optimization of the cloud task scheduling model and algo-rithm is established Task scheduling model for different demands makes the energy efficiency ratio of task scheduling significantly improved in heterogeneous cloud environment.