摘要
提出了一个冗余 TCP堆栈 (RTS)方案以提高 TCP网络连接的可靠性和计算机集群提供服务的可用性 .RTS允许多个服务器同时在一个 TCP连接里工作 ,极大的改善了 TCP连接提供的服务的可信赖性 .在 RTS中还设计了两个算法 (SWMU和 SWFR)来解决在多个 TCP协议堆栈中的同步问题 .实验表明 ,使用 SWFR和 BRC算法可以将网络性能的损耗减到最少 ,并且配备两台服务器的集群所发挥的性能可充分满足在实际使用中所需的高效传送 .
Improving the availability of services of was a key issue for survivabi lity of a cluster system. This paper proposed a scheme called Redundant TCP Stac ks (RTS), providing connection-level availability by exploring the redundant TC P stacks for TCP connections at the server side. Two algorithms (SWMU and SWFR) were proposed to solve the synchronization problem in RTS. The experiment result s showed that the configuration of one primary server with one backup server with S WFR and BRC had acceptable performance to support the server side applications w hile delivering high availability.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第1期6-10,共5页
Journal of Chinese Computer Systems
基金
国家"八六三"项目 (2 0 0 2 AA1Z2 10 2 )资助 .
