基于CloudSim的分类负载均衡调度模型  

Classified Load Balancing Scheduling Model Based on CloudSim

作　　者：李荣荣 牛立栋[1] 孙纪敏[1] Li Rongrong;Niu Lidong;Sun Jimin(The 54th Research Institute of CETC,Shijiazhuang　050081,China)

机构地区：[1]中国电子科技集团公司第五十四研究所,石家庄050081

出　　处：《计算机测量与控制》2018年第3期195-199,共5页Computer Measurement &Control

基　　金：国家自然科学基金(61671179;61504124)

摘　　要：针对传统的集群调度模型效率低下不足以满足用户需求的问题,提出一种基于CloudSim的分类负载均衡调度模型;首先,构建任务请求的指标体系以完成数学模型的建立;接着,采用基于模糊C均值聚类算法的改进算法对请求分类,即用改进的最小支撑树算法获取初始中心,有效性测度获取其分类个数,BP神经网络算法提高其学习能力;然后,采用两次分类的方法对服务器分类,预聚类对服务器进行功能预聚类,模糊关联聚类按处理负载能力对其分类;最后将分类调度模型在CloudSim下仿真实验,综合3种场景,相比于其他调度算法,分类调度模型的最大运行时间最低,资源利用率最高且最高至99%,结果表明该模型更具适应性和高效性,具有工程指导意义。Aiming at the problem that traditional cluster scheduling models are not efficient enough to meet the needs of users,now propose a classified load balancing scheduling model based on CloudSim.Firstly,construct the index system of task request to build the mathematical model;next,use the improved algorithm based on fuzzy C-means clustering algorithm to classify the request,obtaining the number of classification after determining its initial center and improving its learning ability by BP neural network;then,pre-cluster servers according to the function and use the fuzzy clustering algorithm to classify them according to their processing ability.Finally,the classification scheduling model is simulated under CloudSim.By comparing with other scheduling algorithms in three different scenarios,classified model has the shortest maximum running time and the resource utilization up to 99%.The results show that the proposed model is more adaptive and efficient and has engineering significance.

关 键 词：负载均衡 模糊聚类 BP神经网络 CloudSim 调度 

分 类 号：TP193[自动化与计算机技术—控制理论与控制工程]