SQL Server maintains a write-ahead transaction log for each database. This log helps to maintain database updates in cache memory to ensure that data is not written to disk before it has been committed. Database writes occur as a part of the checkpoint procedure.

SQL Server determines the checkpoint frequency based upon the "recovery interval." This interval is a configuration parameter that indicates the maximum time interval that can be tolerated during a system restart. When checkpoint occurs the portion of the transaction log that is no longer needed for system restart becomes inactive and is optionally truncated. The recovery strategy determines whether the transaction log is truncated or not.

If the checkpoint procedure does not truncate the transaction log, then it can be backed up. Then it can be used for point-in-time recovery, failure from disk crash, or move and copy operations.