Definition

Checkpointing is the process of flushing all committed changes from the write-ahead log (WAL) to the main database storage, and then clearing the log.

Core concept

Without checkpoints, the log would grow indefinitely. Periodically (or when the log reaches a threshold), the DB writes all in-memory changes to disk and marks a checkpoint. This means everything before that point is safely on disk, so the log can be truncated.

Use cases

All WAL-based systems do checkpointing (Postgres, SQLite, etc.). Checkpoints shorten recovery time.

Trade-offs

Checkpoint operations can cause write bursts and latency when flushing pages. Tuning checkpoint frequency balances performance vs recovery time.

Example

PostgreSQL’s CHECKPOINT writes all dirty pages to disk. After checkpoint, WAL segments before that point can be archived or deleted.

References

“After a certain number of operations, perform a checkpoint: write all changes in the WAL to the database and clear the log”[[28]](https://en.wikipedia.org/wiki/Write-ahead_logging "Write-ahead logging - Wikipedia").