云计算中基于预先分类的调度优化算法  

Cloud computing scheduling optimization algorithm based on reservation category

作　　者：左利云[1] 左利锋[2] 

机构地区：[1]广东石油化工学院实验教学部,广东茂名525000 [2]郑州宇通客车股份有限公司新能源产品部,河南郑州450016

出　　处：《计算机工程与设计》2012年第4期1357-1361,共5页Computer Engineering and Design

基　　金：广东省科技计划基金项目(2007B010400042);广东省自然科学基金项目(06029274);茂名市科技计划基金项目(20091009)

摘　　要：针对云计算中现有调度算法为追求最短完成时间而不能很好兼顾负载平衡的问题,提出基于预先分类的Min-Min调度算法,该算法先利用能衡量资源计算和通信能力的属性信息对资源进行划分等级,再求出每个调度任务在资源中的最小执行时间,计算任务对应资源等级与最小执行时间的乘积,使用该乘积最小的任务-资源对进行调度。解决了原始Min-Min调度算法负载不均衡的问题,兼顾了执行时间最小和负载均衡。模拟的云仿真系统实验结果表明,该算法在平均任务响应时间、平均任务执行速度下降比和系统利用率等方面优于原始的Min-Min调度算法。In current cloud computing, some scheduling algorithms in order to pursuing the shortest completion time so that igno- ring the load balancing. To solve this problem, an advanced Min-Min algorithm based on pre-classification is proposed. In the new algorithm, the resources classifies through the information of measuring the ability of computing and communications. Then, it is calculated the minimum execution time of each task and calculate the product minimum of the resource level and of the minimum execution time for scheduling. Comparison with the original algorithm, this new algorithm compromises between the minimum execution time and the load balancing. Finally, the cloud simulation experiments with this new algorithm shows a good capability in the average task response time, task execution rate of decline in the average ratio and system utilization relative to the original algorithm.

关 键 词：云计算 预先分类 响应时间 完成时间 速度下降比 

分 类 号：TP392[自动化与计算机技术—计算机应用技术]