Designing Layered High‑Availability Architectures for PostgreSQL on a Budget
summary
Claims a practical framework that composes incremental Postgres HA layers—offsite backups, WAL shipping, hot async standby, sync replica and a cross‑region warm standby—to hit defined RPO/RTO targets on a tight budget (examples include achieving four‑nines availability without enterprise licenses). Uses Patroni for automated failover, Linux watchdog and replication slots, recommends right‑sizing and 5:1 backup compression, warns synchronous replication adds write latency, gives seconds RTO for auto failover and RPO zero with sync in‑region.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:23RPO RTO and failure scope
- 7:04Layered architecture overview
- 13:43Layered architecture designs
- 24:16Split brain and Patroni failover
- 28:13Synchronous replication trade-offs
- 30:19Cost-saving techniques
- 34:09Key takeaways and testing
related talks
46mDeploying Maximum HA Architecture With Postgres
47mPostgreSQL High Availability: The Considerations and the Candidates
55mHigh Availability with PostgreSQL and Pacemaker
50mEvolution Of PostgreSQL Replication and High Availability (upto ~ 99.999%)
38mScaling PostgreSQL Replicas Under Load
57mReliability features in Postgres Pro
42mDon't Do High Availability, Do Right Availability
41mEnsuring High Availability in Self-Managed PostgreSQL with Patroni