【作者】 杨献烽；

【导师】 曹争；

【作者基本信息】 东南大学 ， 计算机系统结构， 2006， 硕士

【摘要】 作为下一代互联网的主要协议,IPv6受到越来越多的关注。国内CERNET2 IPv6主干网的开通,更将IPv6的发展推至一个崭新的阶段。但从IPv4完全过渡到IPv6仍需要相当长的时间,在这漫长的过渡期中,需要纯IPv4站点和纯IPv6站点之间能够直接通信。目前,有很多有用的过渡技术,但绝大多数是关于单播的。组播作为IPv6必须支持的特性之一,同样存在着过渡问题。成功用于单播的协议转换技术,并不完全适用于组播,因为单播采用的是一对一的通信方式,而组播采用一对多的通信方式。国际上已经有相关组织探讨了组播过渡的问题,国内也有少数科研工作者开始关注这方面的研究,但目前都未提供可用于实际网络的具体解决方案。本论文研究了几种具有代表性的现有组播过渡技术,分析了这些技术的工作原理、优缺点和各自的适用性。在此基础上,针对IPv4组播域和IPv6组播域中主机相互进行组播通信的问题,论文提出了一个基于协议转换技术的组播过渡模型——MTG(Multicast Transition Gateway)。MTG是一个网关式系统,它位于IPv4组播网络和IPv6组播网络的边界。MTG针对常用的组播通信需求,支持多种部署方案,采用可管理可扩展的设计原则。MTG针对组播视频会议和组播视频分发两种主要组播应用,利用灵活可靠的跨域(IPv4组播域和IPv6组播域)组播交互机制,较好地解决了一对多、多对多两种组播模型的过渡问题。论文在基于网关技术(网络层的组播协议转换技术)的基础上,设计了简单、清洁、高效、可管理和可扩展的MTG模型。本文根据组播地址映射是否随时间变化,提出了静态组播地址映射和动态组播地址映射的概念。既可以实现等量组地址空间的映射,也能完成大组播地址空间到小组播地址空间的映射。本文在深入分析Linux内核对组播支持的基础上,提出使用可装载的模块编程技术,利用Linux内核防火墙netfilter框架,在内核中网络层报文处理的入口对组播报文进行处理。使MTG具有较高的报文处理性能。同时也可以支持诸如SAP等应用层协议的扩展。在分析了ASM组播模型的基础上,本文研究了组播成员活跃性探测问题,提出了可以使用SNMP采集RP组播路由器中相关MIB和使用可编程组播路由器作为RP相结合的方法,获取组播成员活跃性信息。论文还提出了基于端系统的组播会话应答协议SRP,可以实现组播模型无关的组播成员活跃性探测。本文不仅研究了MTG应用于ASM的情况,还对应用于SSM的情况作了一定的分析。论文还针对不同的网络结构,给出了MTG的网络部署以及与NAT-PT的联合部署。对组播过渡网关模型的功能和性能测试表明,该模型能够有效地支持IPv4和IPv6之间的组播通信,并且具有着良好的可管理性和可扩展性。更多还原

【Abstract】 As the main protocol of the next generation Internet, IPv6 has got more and more attention. The CERNET2 project has made the development of IPv6 to a brand new phase. But IPv4 and IPv6 will co-exist for many years, possibly decades, so finding ways for the two to interoperate is important. There are a number of transition techniques that might be of help. But most of these have focused on unicast.As one of the main features of IPv6, multicast also needs to be considered. The transition technique that works well in unicast may not be fit for multicast, because multicast is one-to-many while unicast is one-to-one. Although a large number of reports and papers are written on transition. But very little has been written on multicast.This paper first studies some typical solutions for the multicast transition, and analyses their mechanism, strongpoint and shortcoming, and applicability. And then, a multicast transition model named MTG based on the protocol translation technique is given. MTG is a gateway placed between IPv4 and IPv6 network. It can be used for multicast video conference and multicast distribution. According to whether the multicast address mapping will change, the paper divides it into two parts: static mapping for address mapping with same size, and dynamic mapping for the others.After analyzing the implement of multicast in Linux kernel, this paper uses loadable module and netfilter framework to make MTG working on the network layer. And this enables MTG’s high performance and expansibility for application layer protocol.This paper studies multicast membership activity for ASM. And it puts forward the method by collecting RP’s MIB through SNMP or using programmable multicast router as RP to get multicast membership activity. And another method named SRP may be useful by placing the function on hosts than router.This paper also discusses network deployments and some SSM issues.The testing results show this model can enable the interoperation between IPv4 and IPv6 multicast, and is manageable and scalable.更多还原