What is "checkpoint and restart" (or "checkpointing") in the context of fault-tolerant computer architecture?

A Periodically saving the complete state of a running computation so that, if a fault occurs, the system can roll back to the most recent checkpoint and resume rather than restarting from the beginning B Checkpointing means permanently deleting all program state at regular intervals C It is a technique used only for saving video game progress, with no relevance to architecture D Checkpoint and restart requires manually rewriting the entire program after every fault