【作者】 赵辉；

【导师】 唐好选； 周国辉；

【作者基本信息】 哈尔滨工业大学 ， 软件工程， 2015， 硕士

【摘要】 在互联网日新月异的发展下,使得许多行业把主要业务转移到网络上开展。数据通信行业作为网络应用最早的行业自然离不开互联网,媒体、广告行业的核心业务对网络依赖程度极高,很多服务行业更是利用网络的便利性发展网上业务,网络还使得零售业发生了前所未有的变革。互联网的大潮下,各行各业都在积极将业务搬到网上,不然就面临的就是倒闭。随着网络带宽指数级的增长趋势,网络上广泛部署了大量事关行业生死的业务,一旦基础网络发生故障,即使是很短时间的,造成的损失也是不容小觑的。本文所设计的系统,正是用作在基础网络瘫痪时,可以快速的恢复正常工作、减少系统瘫痪持续的时间。论文研究了路由器的主备倒换系统发展状况,根据不同的应用场合主备倒换系统的实现有所不同,随着周边技术的发展,也为主备倒换系统的实现提供了多种方案。无论是那种方案其根本思想就是提供冗余设计,本课题采用1+1冗余备份机制,结合HA高可用协议提高路由器的主备倒换系统的可靠性,从而减少故障持续时间。整体系统的开发环境使用C语言,开发工具使用source insight,编译工具使用MIPS CPU架构交叉编译工具链,运行环境为64位路由器系统。本文主要的研究内容分为三大模块,板间通信模块、数据同步模块和主备倒换模块。板间通信模块主要实现了两个通道,数据通道和控制通道,数据通道位于应用层之下、链路层之上,为用户提供可靠的消息传输机制,用户可以使用单播、组播的方式与其他用户通信;控制通道实现一个简单的心跳检测协议,随时监控各个单板的状态,用于快速检测故障板卡并使其恢复工作。数据同步模块主要实现主用板与备用板之间的数据同步,经过热备份主备倒换流程的四个阶段的同步,备用板和主用板数据基本保持一致,当主用板故障备用板随时可以接替主用板继续工作,并且业务不会中断。主备倒换模块,主要实现能够快速有效的监测到主备倒换中断,完成主备板状态切换。测试部分模拟了实际业务中路由器的工作环境作为测试环境,通过对系统进行功能和非功能测试。实验结果表明本系统满足预期的功能点要求和性能要求,在主备倒换过程中能够保证不丢包,有效的处理网络流量,减少故障持续时间。更多还原

【Abstract】 With the rapid development of the Internet, many industries have carried out the main business on the network. As the earliest industry put business on line, data communication naturally cannot be separated from the Internet. The core business of the media and the commercial industry is highly dependent on the network. Many service industries take advantage of the Internet to develop their online business. The Internet has also made unprecedented revolutions to the retail business. Under the tide of the Internet, businessmen are actively moving the business to the Internet, or the bankruptcy will not be far away. With the network bandwidth exponential growth trend, the network has widely deployed a lot of vital business which is considered to life and death about the industry. Even the basic network failure for a short time, the loss is not to be underestimated. The system designed in this paper could quickly resume normal state, and reduce the duration of the system paralysis in basic network failure.This paper studies the situation of the development of Master-slave-switch system. According to different application occasions Master-slave-switch system is different. With the development of the peripheral technology, offers a variety of choices to achieve Master-slave-switch system. Whatever the main idea is redundancy design. In this thesis, “1+1 redundancy backup mechanism” combined with the HA protocol highly improved the reliability of Master-slave-switch router system and reduced the duration of fault. The overall system development environment is C language, the development tool used insight source, the compiler tool is the CPU MIPS structure cross compiler tool chain, the operating environment is 64 bit router system.There are three modules in this theory: inter-board communication module, data synchronization module and the switchover module. The inter-board communication module mainly realizes two channels, data channels and control channels. The data channel is between the application layer and the data link layer, and provides reliable message transmission mechanism for the user. Users can communicate with other clients by unicast and multicast. Control channel is to achieve a simple heartbeat detection protocol to monitor the status of the individual board at real time, and for the rapid detection of the fault board in order to recover normal. Data synchronization module mainly realizes the main synchronization of data between the main board and the standby board, after a hot backup process fo ur stages of synchronization, the backup board and main board data remained the same. Once the main board occurs fault, the standby board can substitute the master board continue work immediately, and the communication will not be interrupted. Master-slave-switch module is mainly to achieve rapid and effective monitoring to the interruption, complete main board and standby board state switch.The testing part simulates the working environment of the router in the actual service and goes through the function and the non-function test of the system. Experimental results show that the system meets the expected functional point and performance requirements: it is no loss during the switch, and can handle network traffic effectively, also can reduce the failure duration.更多还原