一种基于负载熵的层次负载均衡算法

发布时间：2018-03-20 02:13

本文选题：负载均衡　切入点：分布式系统　出处：《计算机应用》2016年S2期 　论文类型：期刊论文

【摘要】：针对分布式集群异构环境下集群节点负载不平衡引起的资源利用率低、作业响应时间长、系统服务质量不佳的问题,根据熵的理论给出集群负载熵的定义,并提出一种基于负载熵的层次负载均衡算法。该算法采用将静态轮询负载和基于负载熵的动态负载相结合的分层负载策略,并将集群划分成若干个均衡域,很好地规避了因集群规模太大引起的节点通信延迟的问题。在Cloudsim上仿真实验表明,该算法相对于Cloudsim自带的先来先服务(FCFS)算法性能提高26.1%,相对于基于并行计算熵的同构集群负载均衡(PCEBLB)算法性能提高12.04%。实验结果表明该算法对节点负载具有良好的均衡性,有效地控制了集群负载失衡的问题,提高了集群系统的资源利用率。
[Abstract]:Aiming at the problems of low resource utilization, long job response time and poor system service quality caused by the load imbalance of cluster nodes in distributed cluster heterogeneous environment, the definition of cluster load entropy is given according to the theory of entropy. A hierarchical load balancing algorithm based on load entropy is proposed, which combines static polling load with dynamic load based on load entropy, and divides the cluster into several balancing domains. The problem of node communication delay caused by the large cluster size is well avoided. The simulation results on Cloudsim show that, Compared with Cloudsim, the performance of this algorithm is improved by 26.1 and 12.04, respectively. The experimental results show that the proposed algorithm has a good load balance for nodes, and the performance of PCEBLB-based isomorphic cluster load balancing algorithm based on parallel computing entropy is improved by 12.04. the experimental results show that the proposed algorithm has a good performance in balancing the load of nodes, and the performance of the proposed algorithm is better than that of the algorithm based on parallel computing entropy. The problem of cluster load imbalance is effectively controlled and the resource utilization of cluster system is improved.
【作者单位】：中国科学院成都计算机应用研究所;西南民族大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(61379019)
【分类号】：TP301.6

【相似文献】