一种基于Nginx的UDP反向代理服务器数据转发策略  

A UDP reverse proxy server data forwarding strategy based on Nginx

在线阅读下载全文

作  者:郭大伟 张伟[1,2] 姜晓艳[1] GUO Dawei;ZHANG Wei;JIANG Xiaoyan(Computer School,Beijing Information Science&Technology University,Beijing 100192,China;Advanced Innovation Center for Materials Genome Engineering,Beijing Information Science&Technology,Beijing 100192,China)

机构地区:[1]北京信息科技大学计算机学院,北京100192 [2]北京信息科技大学北京材料基因工程高精尖创新中心,北京100192

出  处:《北京信息科技大学学报(自然科学版)》2019年第6期87-91,共5页Journal of Beijing Information Science and Technology University

摘  要:为了向互联网视频直播用户提供良好体验,在传统UDP转发模型基础上,面向Nginx设计了一种基于哈希表的UDP数据转发策略,主要是通过基于红黑树的哈希表、复用连接池等技术,为数据包转发过程提供足够数量的socket套接字,避免转发过程的丢包现象,提高UDP数据包转发量,解决了UDP数据包转发时获取不到socket套接字这一瓶颈问题,从而对服务器的性能进行优化。实验结果表明使用优化的模型后,UDP数据包的实时转发数量得到明显提高,可达到百万级别,有效支撑大流量下Nginx对UDP数据包的反向代理,显著提高了基于Nginx的UDP反向代理服务器性能。In order to improve the quality of video live broadcast for better user experience, on the ground of traditional UDP forward model, this paper implements a Nginx-oriented UDP data forwarding strategy based on hash table through the establishment of a hash table based on the red-black tree, and the reuse of connection pool technology for packet forwarding process to provide a sufficient number of sockets. The strategy can avoid packet loss in the forwarding process, improve the amount of UDP packets forwarding, and solve the bottleneck problem of not getting socket in UDP packets forwarding, so as to optimize the performance of the server. The experimental results show that the use of the optimized model can significantly improve the real-time forwarding number of UDP packets to the level of millions, effectively support the reverse proxy of Nginx to UDP packets under large traffic, and significantly improve the performance of the UDP reverse proxy server based on Nginx.

关 键 词:网络时延 UDP 反向代理 哈希表 连接池 红黑树 

分 类 号:TP202[自动化与计算机技术—检测技术与自动化装置]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象