Database Systems Practice Test

Checkpointing marks a point in the recovery process where the system flushes all dirty pages to disk and records that state. By writing updated pages from memory to disk and logging this event, the database creates a known-good starting point for crash recovery. When the system restarts, recovery uses the most recent checkpoint plus the log to redo actions from that point and to undo any transactions that didn’t commit before the crash. The recovery log itself is the record of changes used during recovery, but it is not the moment updates are written to disk. A commit signals that a transaction’s results are durable, and rollback undoes uncommitted work; neither inherently defines the disk-write checkpoint. Thus, this term correctly identifies the moment in the recovery process when disk writes occur.