CHECKPOINTING AND ROLLBACK RECOVERY FOR WINDOWS NT APPLICATIONS

In this paper a checkpointing and rollback recovery （CRR） mechanism for Windows NT applications is introduced and a checkpointing and recovery tool, WinNTCkpt is presented. WinNTCkpt can be used transparently to checkpoint and recover applications on Windows NT environment. Dynamic function wrapper and standard Win32 APIs are employed to implement this tool. WinNTCkpt is also the pivotal part of the high availability parallel computing environment under research and development. Furthermore, CRR is the technical basis for process migration to implement load balance and improve system availability on Windows NT.