The journey towards active-active replication in PostgreSQL
summary
PostgreSQL moves from single-primary setups toward multi-primary capability by extending logical replication: Postgres 16 adds bidirectional replication via origin filtering (enabling table-to-table replication between two instances), allows logical replication from standby to offload the primary, and introduces parallel apply of large transactions (thresholds around ~75KB) to avoid WAL build-up. Other concrete changes: binary-format initialization speeds warm-up, PG_CREATE_SUBSCRIPTION lets non-superusers own subscriptions, and replica‑identity‑full can use btree indexes. DDL replication, robust conflict detection/resolution (node sync and last‑commit‑wins), and sequence access methods remain unfinished.
AI-written summary · about
chapters (8)
- 0:00Introduction to Replication
- 1:04Types of Replication
- 17:36Postgres Development Timeline
- 18:50Replication Flexibility
- 19:37Deep Dive into New Features
- 33:06Collaborative Development
- 34:23Future Roadmap
- 43:06Conclusion and Q&A
related talks
53mThe journey towards active replication in PostgreSQL
46mPostgreSQL 16 and beyond
50mSparta’s Dual Kingship and PostgreSQL Active-Active
31mPast, Present, and Future: Logical Decoding and Replication in PostgreSQL
39mLogical Replication handling of large transactions
59mLogical Replication...LIVE!
24mPostgreSQL physical replication - internals, latest development and opportunities
49mPostgreSQL Replication - The Most Exciting Technology on Earth
