Past, Present, and Future: Logical Decoding and Replication in PostgreSQL
summary
Starts with the logical decoding API (Postgres 9.4) plus replication slots and replica identity, extended by pglogical to add publish/subscribe; Postgres 10 brings native CREATE PUBLICATION/CREATE SUBSCRIPTION (PG10); PG15 adds row and column filtering and PG16 adds parallel apply and binary initial sync that cut initial sync time and steady‑state lag dramatically; PG17 preserves slots across failover; PG18 adds DDL replication, conflict detection and pg_stat_subscription_stats; common stacks use pgoutput or wal2json → Debezium → Kafka → analytics for zero‑downtime upgrades and CDC, while BDR/Spock/pgEdge pursue multi‑master.
AI-written summary · about
chapters (15)
- 0:00Music & introduction
- 0:40Speaker background and experience
- 1:30Agenda: past, present, future roadmap
- 2:34Physical vs logical replication explained
- 4:27Inside logical decoding architecture
- 7:59Origins in Postgres 9.4 and pglogical
- 10:45Native replication evolution to PG18
- 14:17Key features: filtering and parallel apply
- 17:14High availability, failover, and conflict detection
- 19:34Native vs extension-based replication tradeoffs
- 21:18CDC pipelines and real-world use cases
- 25:12Real-world patterns of logical replication
- 26:42Performance tuning and common pitfalls
- 27:55Future: multi-master distributed Postgres
- 30:18Key takeaways and closing remarks
related talks
53mThe journey towards active replication in PostgreSQL
54mUnderstanding Logical Decoding and Replication
20mKeynote
44mThe journey towards active-active replication in PostgreSQL
42mlesser known Superpower of Postgres Logical Replication
59mLogical Replication...LIVE!
50mPGLogical the logical replication for Postgresql
46mPostgreSQL 16 and beyond