基于优先排队论网络延迟云计算资源调度算法  

Resource scheduling algorithm based on priority queuing and network delay in cloud computation

作　　者：崔建明[1] 刘佳祎[2] 杨呈永[1] 

机构地区：[1]桂林理工大学现代教育技术中心,广西桂林541000 [2]桂林理工大学信息科学与工程学院,广西桂林541000

出　　处：《桂林理工大学学报》2017年第2期360-365,共6页Journal of Guilin University of Technology

基　　金：广西高等学校科研项目(YB2014149)

摘　　要：针对虚拟机在实际应用环境中,对于不同数据在不同情况下需要不同优先级输出的问题,采用运筹学优先制M/M/1排队模型,对虚拟机请求作出网络延迟分析并对传统的顺序输出方法加以改进。结合数据资源在云计算环境下的Map-Reduce模型,提出新的调度算法以及不同于传统算法的度量指标。经过Cloud Sim仿真软件进行模拟实验,结果表明,新的算法将网络延迟以及物理机和虚拟机的CPU、内存等资源综合考虑,在性能上要优于随机算法、转轮算法等传统算法,大大改善了负载均衡度,缩短任务调度总时间,并使总调度时间的负载效率得以提高。In virtual machine application,as different data under different circumstances require different priority output,from the operation research system of M/M/1 queue model,network latency analysis to virtual machine request output method is improved and so the traditional order. The combinations of resources in cloud computing environment Map-Reduce model put forward a new scheduling algorithm,different from traditional metrics of the algorithm. After Cloudsim simulation software simulation experiment,the results prove that the new algorithm synthesized considers network latency as well as physical and virtual machine CPU,memory and other resources into account. Its performance is better than traditional algorithms,such as random algorithm,wheel algorithm. It greatly improves the balancing degree,speeds up the task scheduling time and the loading time.

关 键 词：云计算 网络延迟 优先制M/M/1排队模型 资源调度 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]