基于负载均衡的Min-Min任务调度算法优化  

Optimization of Min-Min task scheduling algorithm based on load balance

作　　者：王文豪[1] 严云洋[1] 周静波[1] 

机构地区：[1]淮阴工学院计算机工程学院,江苏淮安223003

出　　处：《南京理工大学学报》2015年第4期398-404,共7页Journal of Nanjing University of Science and Technology

基　　金：江苏省科技支撑计划资助项目(BE2012112);江苏省高校自然科学研究项目(14KJB520006);淮安市科技计划项目(HAG2013068)

摘　　要：为了解决Min-Min调度算法中存在的负载不平衡问题,提高集群系统的负载均衡性,该文提出了一种基于Min-Min极限下压算法的负载模糊分类与局部重调度算法(Load fuzzy classification and local re-schedule algorithm,LFC-LRA)。引入模糊分类的思想,根据各节点的负载大小,将节点分成三种类型:重负载、中负载和轻负载;对负载较重和较轻的节点进行重新调度,使用Min-Min极限下压算法压缩这些节点的任务完成时间,改善算法的负载失衡问题。实验结果表明:改进后的算法具有较好的负载均衡性,能有效地提高资源的利用率,降低系统的任务完成时间。In order to solve the load imbalance problem in Min-Min scheduling algorithm and improve the load balance of cluster systems,this paper presents a load fuzzy classification and local reschedule algorithm( LFC-LRA) based on Min-Min limit down compression algorithms. A fuzzy classification theory is introduced in this algorithm. According to the load of each node,nodes are divided into three types,heavy load,medium load and light load. Nodes with heavy load and light load are rescheduled and their task completion time is further compressed by Min-Min limit down compression algorithms which improve the load imbalance problem of the algorithm. The experimental results showthat the improved algorithm can achieve a better load balance,significantly improving the system utilization,and reducing the system task completion time.

关 键 词：集群 任务调度 算法优化 MIN-MIN算法 模糊分类 局部重调度 负载均衡 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]