point-in-time recovery
Restoring a database to its exact state at a chosen moment by loading a backup and replaying logged changes up to that moment.
In more detail
Point-in-time recovery, or PITR, starts from the last full backup and then replays the transaction log up to just before the failure or mistake. It can close the gap between nightly backups to minutes, which helps meet a tight recovery point objective. It is also how you undo a bad change, by stopping the replay just before it happened.
Example
Restore last night's full backup, then replay the transaction log to 14:36.Related terms
- journaling — Recording each change in a log before applying it, so a system can recover cleanly or replay changes after a failure.
- RPO — Recovery point objective, the most data a business can afford to lose in an outage, measured as the time before the failure.
- full backup — A backup that copies all of the selected data every time, whatever has changed since the last one.
- backup interval — How often backups are taken, such as hourly or nightly, which sets the most data you could lose between one backup and the next.
Taught in
- CloudFormation In Depth: Replace Or Update? — AWS Certified DevOps Engineer – Professional (DOP-C02) Exam Prep
- Backup Architectures: 3-2-1, Immutability & Air-Gaps — Cybersecurity Fundamentals, GRC & Cryptography
- Internal Audits, Sampling & Evidence Chains — Cybersecurity Fundamentals, GRC & Cryptography
- Running Storage And Data — Google Cloud Associate Cloud Engineer Exam Prep
References
- PostgreSQL documentation: Continuous Archiving and Point-in-Time Recovery
- Amazon RDS User Guide: Restoring a DB instance to a specified time