一种基于Active Replication的TCP连接容错技术  

A Fault-Tolerant Schema for TCP Connection Based on Active Replication

作　　者：程斌[1] 金海[1] 徐婕[1] 

机构地区：[1]华中科技大学计算机科学与技术学院,湖北武汉430074

出　　处：《计算机工程与科学》2005年第9期71-73,共3页Computer Engineering & Science

基　　金：国家863计划资助项目(2002AA1Z2102)

摘　　要：本文提出一种基于ActiveReplication的TCP连接容错技术(ARTCP),采用两层的Cluster结构,并通过前端的分发器和聚合网关机维持多个同步的TCP连接副本,达到了相互容错的效果。测试结果表明,该技术不仅保证网络服务连接级的高可用性,而且具有恢复时间短、开销小的优势。It is an effective way to mask the faults of active TCP connection by providing high availability network services. This paper proposes a new fault- tolerant scheme for TCP connection, which is based on active replication. The technique uses a dispatch server and an aggregate server to synchronize multi-images for the same active TCP connection with the clients in order to mask the faults of the TCP connection. The experimental results show that the technique not only provides high availability of network services, but also has less recovery time and lower overhead.

关 键 词：TCP 容错 ACTIVE REPLICATION 同步 

分 类 号：TP302.8[自动化与计算机技术—计算机系统结构]