摘要
提出了一种基于软件定义网络的分布式数据库负载均衡算法,将数据、控制、应用分离的同时计算服务器集群中单个服务器的实际负载。通过查询流量采样记录来决策最少连接的服务器路径,减少了访问请求的响应时间,提高了系统吞吐量和容错能力,实现了充分利用服务器资源的目的。内网中的分布式数据库实验对通用负载均衡技术和基于软件定义网络的负载均衡技术进行了比较,在不同服务器集群的负载状态下,后者的平均响应时间小于前者,并得到了更好的负载均衡效果。
A distributed database load balancing algorithm based on software defined networking was proposed, which separated the data, control and application to calculate the actual load of a single server in the server cluster. By querying the streams sampling record, the minimum number of server paths were determined, the request re- sponse time was reduced, the system throughput and fault tolerance were improved, and the full use of the purpose of server resources was achieved. The general load balancing technology was compared with the load balancing tech- nology based on the software defined networking in the internal network of the distributed database experiment, the result shows that in the load state of the different server clusters, the average response time of the latter is smaller than that of the former, and can get better load balancing effect.
出处
《电信科学》
北大核心
2017年第9期190-196,共7页
Telecommunications Science
基金
国家电网公司科技基金资助项目(No.SGBJDK00KJJS1500180)
国网信息通信产业集团有限公司科技基金资助项目(No.SGITG-KJ-JSKF[2015]0010)~~
关键词
可扩展负载均衡策略
软件定义网络
自适应流量采样
分布式服务器集群
extensible load balancing strategy, software defined networking, adaptive stream sampling, distributed server cluster