基于负载反馈的分布式数字集群动态负载均衡算法  

Dynamic load balancing algorithm based on load feedback for distributed digital trunk

作　　者：曲乾聪 王俊[1] Qu Qiancong;Wang Jun(School of Electronic&Information Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学电子信息工程学院,北京100044

出　　处：《计算机应用研究》2022年第2期526-530,542,共6页Application Research of Computers

基　　金：国家重点研发计划资助项目(2018YFB1601504)。

摘　　要：针对传统负载均衡算法不能满足公网数字集群系统高并发用户请求和快速呼叫建立等需求,提出一种基于负载反馈的分布式数字集群动态负载均衡算法,实现公网数字集群系统负载均衡,提高用户容量。首先建立参与MCPTT服务器的静态负载和动态负载监控机制和指标;然后利用加权轮询算法为用户分配参与MCPTT服务器,并通过用户请求的处理获得复合负载参数;根据负载指标的反馈更新参与MCPTT服务器权值以动态调整服务器负载。仿真结果表明,该算法的负载均衡效果优于传统算法和其他动态反馈算法,负载均衡度更小、用户请求响应延迟更低。Traditional load balancing algorithms can’t meet the demand of high concurrent requests and quick calling of public network digital trunk.This paper proposed a dynamic load balancing algorithm based on load feedback for distributed digital trunk to achieve load balancing and increase user capacity.Firstly,this paper established monitoring method and monitoring indexes for static and dynamic load of participating MCPTT server.Then,it used weighted round robin algorithm to allocate participating MCPTT server to users and got complex load parameters by dealing user requests.Finally,it updated the weight of participating MCPTT server based on load feedback to dynamically adjust server load.Simulation results show that the proposed algorithm has better load balancing effect,smaller load balancing degree and lower response delay compared with traditional algorithms and other dynamic feedback algorithms.

关 键 词：分布式数字集群 动态负载均衡 动态权值 负载反馈 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]