一种基于QoS的多任务实时集群系统负载均衡算法  

A Load Balancing Algorithm Based on QoS for Multi-task Real-time Cluster Systems

作　　者：白雪飞[1] 黄本雄[1] 

机构地区：[1]华中科技大学电信系,湖北武汉430074

出　　处：《电讯技术》2004年第6期154-157,共4页Telecommunication Engineering

摘　　要：负载均衡算法是多任务实时集群系统的一个关键技术。在分析了常见的轮转式均衡调度算法和任务最少优先法的优缺点基础上,针对实时集群系统处理多类任务的特点,引入了加权负载率的概念来更准确地描述各分布式集群节点的负载状况。提出了通过任务QoS映射为负载权值的思想,并给出了一种改进的任务最少优先算法———加权负载率最小法,描述了以加权负载率表为核心的多任务均衡调度算法。最后,通过仿真验证了该算法明显优于前两种算法。Load balancing is a key problem in a multi-task real-time cluster system. After analyzing the merits and demerits of round-robin algorithm and least tasks first algorithm, this paper introduces LW(load weigh) to describe the load state of cluster node more accurately by mapping the QoS of tasks to weight. Further more, an effective load balancing algorithm and task schedule algorithm are given. Simulation results show the algorithm is prior to the above-mentioned two algorithms.

关 键 词：多任务实时系统 负载均衡 分布式系统 集群 服务质量 算法 

分 类 号：TN915[电子电信—通信与信息系统]