基于模糊聚类的云任务调度算法  

Cloud scheduling algorithm based on fuzzy clustering

作　　者：李文娟[1,2] 张启飞[1] 平玲娣[1] 潘雪增[1] 

机构地区：[1]浙江大学计算机学院,浙江杭州310018 [2]杭州师范大学钱江学院,浙江杭州310012

出　　处：《通信学报》2012年第3期146-154,共9页Journal on Communications

基　　金：浙江省科技厅公益技术应用研究基金资助项目(2010C31003)~~

摘　　要：为了更好地实践云计算提供廉价按需服务的宗旨,提出了一种在模糊聚类基础上,基于两级调度模式的任务调度(FCTLBS,fuzzy clustering and two level based task scheduling)算法,新算法设置用户调度和任务调度2个等级。对资源进行性能模糊聚类;根据任务参数计算资源偏好,使不同偏好任务在不同聚类中选择,缩小了选择范围,更好地反映了任务需求。仿真实验表明,本算法较之同类算法具备一定的优越性。In order to better serve cloud＇s objective of providing low-cost and on-demand services,a two-level and fuzzy clustering based cloud scheduling algorithm named FCTLBS（fuzzy clustering and two level based task scheduling algo-rithm） was proposed.Scheduling was divided into two levels： user scheduling and task scheduling.It used fuzzy cluster-ing method to cluster resources according to their capabilities and computed tasks＇ resource bias coefficient according to their parameters.Thus tasks with different resource preference were able to choose different capability resources in the corresponding clusters.The new algorithm reduces the scale of resource choice,better reflects cloud tasks＇ service re-quirements.Simulation results show that the proposed algorithm has some advantages compared to other kindred algo-rithms.

关 键 词：云计算 模糊聚类 任务调度 分配公平 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]