摘要
提出了一种可伸缩的TMR容错计算系统结构,根据TMR系统出现故障的情况,详细研究了其故障恢复模型和恢复策略。通过综合采用向前和向后恢复方法有效减少了由于实现容错功能而对系统运行进程完成时间的推延并进行了定量分析和验证。
The system architecture of a kind of scalable fault-tolerant computer with triple module redundancy is pro- posed. Detailed study on its fault recovery model and scheme corresponding to different fault occurrence situations is done. Through synthetically adopting forward recovery and roll-back recovery, the impact on the completion time delay of system processes due to the demand on fault tolerance can be efficiently cut down, which is quantitatively analyzed and validated.
出处
《计算机科学》
CSCD
北大核心
2008年第7期119-121,132,共4页
Computer Science
基金
国防重点预先研究项目(413160201)资助
关键词
向前恢复
容错计算
故障恢复模型和策略
系统性能
Forward recovery,Fault-tolerant computing, Fault recovery model and scheme,System performance