一种基于负载熵的层次负载均衡算法

Hierarchical load balancing algorithm based on load entropy

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。

To solve the low utilization rate of resources, long response time and poor Quality of Service （QoS） caused by load imbalance of cluster nodes, a hierarchical load balancing algorithm based on load entropy was proposed. The algorithm fused a static load balancing algorithm round-robin and a dynamic load balancing algorithm based on load entropy. In this algorithm, cluster nodes were divided into several regions of balancing which avoided the problem of high communication cost between nodes in large scale cluster. In the Cloudsim simulation experiments, the proposed algorithm achieved a better balanced workloads. The performance of cluster was improved by 26. 1% and 12.04% compared with First Come First Service （FCFS） algorithm of Cloudsim and Parallel Computing Entropy Based Load Balancing （PCEBLB） algorithm respectively. The simulation results show that the proposed algorithm can control the problem of load imbalance effectively and improve the resource utilization efficiency of the cluster system.