基于虚拟文件操作的文件检查点设置

实现分布/并行系统容错的基础是单进程检查点设置和卷回恢复技术,而对活动文件信息进行保存和恢复则是这种技术的重要方面.提出一种虚拟文件操作策略,实现了对用户文件的检查点设置,有效地解决了发生故障时用户文件内容与进程全局状态的不一致的问题.该方法通过文件块式管理、检查点分布操作等技术,使得在空间开销、正常运行时间、恢复时间等性能指标上优于其他方法,并且具有对用户透明、可最大限度地保留已完成工作的特点.

A Files Checkpointing Approach Based on Virtual File Operations

Checkpointing and rollback recovery of Unix process are the underlying technique of fault tolerance for distributed system and parallel environment. To save and restore the state and the content of active file of the process is an important aspect of checkpointing and rollback recovery. A new file checkpointing approach called virtual file operation (VFO) is presented. VFO buffers all the write operations after a checkpoint until the next one, making all the operations between two checkpoints atomic as a whole.By step-to-step checkpointing,and managing file as blocks,this approach achieves lower overhead than the others.