一种自适应的网络代理集群负载均衡策略

Self-adaptive Load Balancing Strategy in Web Proxy Cluster

现有动态负载均衡策略难以有效应用于任务资源需求特征动态多变、资源需求信息难以准确获取的网络代理集群中,为解决该问题,提出一种适用于网络代理集群的自适应负载均衡策略。该策略的负载模型可以随任务资源需求特征的变化自适应地调整,能够更准确地表征系统的负载状态和节点间的负载差异。实验结果表明,使用该自适应负载均衡策略的网络代理集群的平均响应时间比其他负载均衡策略低15%～20%。

Since the feature of resource requirement in Web proxy cluster is dynamic and resource requirement information is also hard to obtain, existing dynamic load balancing strategies are difficult to be applied in Web proxy cluster. A new self-adaptive load balancing strategy for Web proxy cluster is presented. The load model of proposed strategy can adapt itself to variation of task resource requirement feature dynamically. Therefore load status of system and load difference among nodes can be reflected more precisely. Experimental results show that the average response time of the presented strategy is about 15% to 20% lower than that of other load balancing strategies.