【摘要】 针对敏感行业中分布式应用的容错需求问题,分析介绍Agent、多Agent系统和容错中间件技术,根据Agent和中间件特性结构上的相似性,对利用多Agent技术构建容错中间件作了尝试,并着重研究了失效检测与恢复系统;建立局部检测与全局检测互相结合的双层失效检测模型,提出融入定点恢复和异机恢复的改进型REDO失效恢复策略;最后给出基于JADE的一个系统实现。实验结果显示双层检测模型和改进型REDO恢复策略是可行的、高效率的。更多还原

【Abstract】 Considering the fault tolerance requirements of distributed application in the sensitive industries,this paper analyzes Agent,multi-Agent system and fault-tolerant middleware technology.According to the similarity of Agent and middleware,this paper tries to construct fault-tolerant middleware based on multi-Agent,and focuses on the failure detection and recovery system.A twolayer failure detection model with local detector and global detector is established,and an improved REDO recovery strategy is presented.Finally,an implementation of the system based on JADE is given.The results show that the two-layer failure detection model and the improved REDO strategy are feasible and efficient.更多还原