Design and implementation of a mechanism to recover from failed network processes
一种网络进程故障恢复机制的设计与实现

Based on the commercial PC/Linux platform,a new method combining rollback fault recovery technique with checkpointing technique was put forward and validated to resume failed server network processes when working in C/S model.The method was able to guarantee the consistency and persistency of connections of network communication,i.e.,to guarantee that there are no stops and no stalls with network communications.The method can be used in permanent fault recovery,cluster system,load balancing field,etc.