Deploying Maximum HA Architecture With Postgres
summary
Claims that Postgres 9.x supports 99.99% availability translate into a concrete architecture: a floating VIP, hybrid streaming replication + WAL shipping for fast catchup, manual failover to minimize surprises, backups taken from standbys and tested daily, monthly pg_dump and OmniOS/GFS snapshots, and tooling like PG extractor and PG reorg to track schema changes and remove bloat online. Reported outcomes include sub‑minute switchovers for server failure and the ability to pause WAL apply for about an hour to undo accidental deletes, showing the approach works in practice.
AI-written summary · about
chapters (10)
- 0:00Start
- 0:57Agenda
- 1:59Cap Theorem
- 9:44Goals for Maximum Architecture
- 10:20How Do You Manage Failover
- 12:42Reporting Queries
- 13:57Unplanned Outages
- 22:37Regular Recovery Testing
- 24:15Monthly Pg Dump Backups
- 35:49What the Performance Impact When You Do Pg Reorg
related talks
36mDesigning Layered High‑Availability Architectures for PostgreSQL on a Budget
50mEvolution Of PostgreSQL Replication and High Availability (upto ~ 99.999%)
47mPostgreSQL High Availability: The Considerations and the Candidates
13mBuilding reliable infrastructure with Postgres
57mReliability features in Postgres Pro
22mSecuring Postgres with streaming replication
50mMoving from MSSQL to pgSQL – High Availability and Disaster Recovery
51mScaling Postgres to the next level at OpenAI