Levels of Checkpointing

System Level

The operating system provides the ability for saving a snapshot of the application's memory image to a file for later restart. Available for certain operating systems only. (eg Irix)

Application Level

Certain batch queueing systems provide checkpointing libraries -- these replace parts of the compiler's runtime system to save a memory image snapshot of the process. LSF has this functionality.

User Level Checkpointing

User explicitly provides checkpoint/restart code as part of the application's code