In the instance of the PROD database, the checkpoint (CKPT) process runs after every minute. A database user updates the rows of the ORDERS table. Because of the configuration, the CKPT process gets…

When a checkpoint fires, CKPT instructs the Database Writer process (DBWn) to flush dirty blocks - including those modified by uncommitted transactions - from the buffer cache to their corresponding data files. Consistency is preserved because the redo log captures the change vectors needed to redo committed work, and undo segments contain the before-images needed to roll back uncommitted changes after instance recovery.

Temp files store temporary segment data used for sort and hash operations; checkpoint writes target data files, not temp files.

The control file records structural metadata such as checkpoint SCN and datafile status; modified data blocks are never written to the control file.

Redo log files capture change vectors written by the Log Writer (LGWR) process continuously; CKPT does not write data blocks to redo logs.

Archived redo logs are offline copies of filled online redo log groups used for media recovery; checkpoint activity does not direct data blocks to them.

Before a checkpoint the dirty blocks reside in the buffer cache, but the checkpoint specifically exists to trigger DBWn to write them to disk - they are not simply retained after CKPT fires.