摘要
针对动态反馈算法中反馈过程可能造成过多额外负载的问题,提出一种基于反馈周期的网络负载均衡算法。周期性地向后端服务器收集信息,对节点的实时负载进行评估,结合节点初始性能,以负载容余平衡为目标指导请求分配;建立关于反馈周期的BP网络,以集群的利用率和系统的负载失衡度为输入,对信息反馈周期进行动态调整。用优化网络工程工具进行仿真,仿真结果表明,该算法有效减少了平均响应时间,提高了系统的性能。
The feedback process of dynamic feedback algorithm may cause too much extra load,a network load balancing algorithm based on feedback loop was proposed.Information was collect from backend server periodically,the real-time load of node was evaluated,combined with the node initial performance,aiming at load capacity balance,distribution was requested.The BP networks for feedback cycle were established,and the cluster utilization and system load unbalance degree were taken as inputs,information feedback cycle was dynamically adjusted.Optimized network engineering tool was used for simulation.Results show that the algorithm is effective on reducing the average response time,the performance of the system is improved.
出处
《计算机工程与设计》
北大核心
2016年第4期841-846,共6页
Computer Engineering and Design
基金
广西自然科学基金项目(2013GXNSFBA019282)
广西高等学校科研基金项目(KY2015YB254)
国家级大学生创新创业训练计划基金项目(201410605055)
广西自治区级大学生创新创业基金项目(201410605055
201410605056
201410605057)
关键词
服务器集群
负载均衡
反馈周期
BP网络
server cluster
load balance
feedback cycle
BP neural networks
