Implementing strong cross-platform resiliency in agile environments with logical r

on YouTube ↗33m

" summary

Requirement enforces RPO 15 minutes for ~60 critical apps so physical replication outside the cloud is impossible, backups and WAL access are blocked. Solution uses logical replication per schema across 85 Postgres servers (600 DBs, 1,150 schemas), one subscriber per publisher, automated suspend/resume API wired into CI pipelines, catalog diffing to detect DDL divergence, and full reinitialization (initial sync of ~50–100 GB) when schemas diverge. Monitoring and alerting handle apply errors; approach meets the RPO but is brittle and not fully scalable.

AI-written summary · about

" chapters (9)

1. 0:00Introduction and speaker intro
2. 1:50Context and cloud migration
3. 5:03RPO5 requirement and constraints
4. 8:16Evaluating replication solutions
5. 10:46Why logical replication
6. 13:59Monitoring and handling replication breaks
7. 18:02Schema changes and replication suspension
8. 22:41Results and limitations
9. 24:10Future outlook and Q&A

" related talks

• 36mDesigning Layered High‑Availability Architectures for PostgreSQL on a BudgetUmair Shahid · SREcon26 Americas
• 55mHigh Availability with PostgreSQL and PacemakerShaun M. Thomas · Postgres Open 2012
• 22mPostgreSQL Disaster Recovery FundamentalsRilwan Musa · Postgres Conference Orlando 2025
• 46mDeploying Maximum HA Architecture With PostgresDenish Patel · Postgres Open 2012
• 01h 41mPostgreSQL Upgrades with Logical Replication & Minimal DowntimeIldefonso C · Postgres Conference Orlando 2025
• 50mEvolution Of PostgreSQL Replication and High Availability (upto ~ 99.999%)Stacy Scoggins · PGConf NYC 2021
• 50mMoving from MSSQL to pgSQL – High Availability and Disaster RecoveryRick Lowe · Postgres Conference Orlando 2025
• 51mSpeeding up logical replication setupEuler Taveira · PostgreSQL Conference Europe 2024