← Advanced DBMS learning pathAdvanced DBMS · Module 6 of 10
Recovery and logging
Restore databases after failures.
📖 Key ideas
- Failures: transaction, system crash, media failure.
- Write-Ahead Logging: log changes before writing data.
- Undo uncommitted changes; redo committed ones.
- Checkpoints reduce recovery time.
- ARIES is a well-known recovery algorithm.
💡 Real-life example: The log is a diary of every change: after a power cut, re-read it to rebuild exactly what happened.
WALWrite-Ahead Logging
UndoReverse uncommitted changes
RedoReapply committed changes
CheckpointPoint that limits recovery work