云系统中能量感知任务的整合策略及调度方法

Energy-aware task integration strategy and scheduling method in cloud system

针对目前云系统中虚拟集群间任务调度复杂、能源消耗高和系统成本高的问题,本文提出一种分配CPU利用率和管理虚拟集群间任务的ETI方法,该方法通过限制CPU利用率低于某个特定的阈值实现资源的负载均衡和最小化能量消耗,同时考虑当任务迁移到虚拟集群时能源成本模型中的网络带宽。为了评价ETI方法的性能,将其与改进的经典Min-Min算法和虚拟机部署算法ACOSA比较,在云系统模型中进行实验验证的结果表明:ETI方法能够提高云系统工作稳定性,减少系统能量消耗,比原系统节能约18%,系统成本减少约9%。

To solve the problems of complex task scheduling,high energy consumption and high system cost in the current cloud system,this paper proposes an ETI method to allocate CPU utilization and manage task integration between virtual clusters. It implements resource load balancing and minimizes energy consumption by limiting CPU utilization below a specific threshold,while taking into account the network bandwidth in the energy cost model when tasks migrate to virtual clusters. In order to evaluate the performance of ETI method,it is compared with the improved classical min-min algorithm and virtual machine deployment algorithm ACOSA,and verified by many examples in the cloud system model. The experimental results show that the ETI method can improve the working stability of the cloud system,reduce the energy consumption of the system,save about 18% energy and reduce the system cost by about 9% compared with the original system.