面向用户体验的动态负载均衡算法研究  

Research on a dynamic load balancing algorithm based on quality of user experience

作　　者：郑晓辉 史骁 金岩[1] 宋永浩 唐宏伟[1] 赵晓芳[1] Zheng Xiaohui;Shi Xiao;Jin Yan;Song Yonghao;Tang Hongwei;Zhao Xiaofang(Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190;University of Chinese Academy of Sciences,Beijing 100049)

机构地区：[1]中国科学院计算技术研究所,北京100190 [2]中国科学院大学,北京100049

出　　处：《高技术通讯》2021年第4期359-366,共8页Chinese High Technology Letters

基　　金：国家重点研发计划(2018YFB0904503)资助项目。

摘　　要：随着互联网用户数量的迅速增长,Web服务面临着巨大的访问压力,对Web服务器持续提供服务提出了更高的要求。当负载过重时可能会导致服务器宕机等严重状况,从而影响Web服务器对用户HTTP请求的响应,进而影响服务质量,因此Web服务器集群需要更好地响应用户请求。本文提出了一种基于用户体验的动态负载均衡算法(Q RFS),在保证用户满意度的前提下,兼顾服务器的负载和请求调度的公平性。实验表明,QRFS算法能够获得更好的用户满意度。此外,在请求的平均响应时间上,QRFS算法较加权轮询(WRR)算法提高了12%,较加权最小连接数(WLC)算法提高了11%;在服务器负载的均衡性上,QRFS算法取得了更好的效果,节点间的负载方差更小。With the development of information technology and exploding of the number of Internet users,Web services are facing huge access pressure,which puts forward higher requirements for Web servers to provide services continuously.Web server overloading can lead to serious conditions such as server downtime,which can affect the response to the users’HTTP requests and thus impacting the quality of experience.A quality of experience-based request fair scheduling(QRFS)algorithm is proposed,which gives consideration of different types of requests and users’priority,with the goal of ensuring user satisfaction and achieving good fairness and requests distribution among server nodes.Experiments show that QRFS achieves better user satisfaction than the weighted round robin(WRR)algorithm and the weighted least connection(WLC)algorithm.Besides,QRFS improves the average response time by 12%compared to WRR,and 11%compared to WLC.In terms of load balance of the server,QRFS algorithm achieves smaller load variance.

关 键 词：负载均衡 WEB服务器集群 公平性 调度 用户满意度 平均响应时间 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]