Research And Implementation On High-Available Networks And BGP Graceful Restart

With the rapid development of mission-critical network applications, more research is devoted to the implement of high-available networks which is one of hot discussed problems of networks field by the demand of market. To achieve the high-available criterion, all components in networks, including hardware, software, route protocol etc., should be contribute to the goals. Designing of highly available systems requires an appropriate balance of component redundancy, software recoverability and failover mechanisms, service-level expectations, and system cost and complexity.BGP is an inter-domain routing protocol used uniquely in Internet between ASs. It is typically deployed at the WAN edge between the domains of different network. Its stability and availability will affect the performance of network applications inevitably. The IETF has developed a set of routing protocol extensions about Graceful Restart mechanism to improve the availability of whole networks.In this paper, based on the study of BGP protocol and the related protocol extensions which support IPv6, we introduce BGP Graceful Restart mechanism in detail. We lay emphasis on the discussion about the problems with legacy router architectures, with a focus on the limitations of control plane recovery mechanisms. In addition, we outlines the fundamental architectural changes necessary to achieve the "five-nines" availability standard for IP networks, and essential attributes for a high-availability architecture. The effect of all components in networks on the high-availability implement is studied. The scheme of implement of BGP Graceful Restart on the IPv6 router which is based on the forwarding and control separation framework is presented. The way of extending forwarding and control separation ability and BGP routing protocol is discussed at last.