New feature
Public Preview
PostgreSQL connector
PostgreSQL connector is a Databricks data engineering capability, introduced December 2025.
A managed Lakeflow Connect database connector that incrementally ingests PostgreSQL 13+ tables (RDS, Aurora, Azure, Cloud SQL, on-premises) into Databricks using logical replication and an ingestion gateway.
- It handles the full PostgreSQL date range, including BC dates, using astronomical year numbering - so 1 BC lands in Databricks as year 0 and 2 BC as year -1.
- Deleting a pipeline does not delete its replication slot, so a forgotten slot keeps holding WAL on the source until someone cleans it up by hand.
- Money is just text to it - MONEY columns arrive as strings, along with TIME, TIMETZ and plain TIMESTAMP values.
Limitations: PostgreSQL 13 or later and primary instances only, username and password authentication only, the ingestion gateway must run continuously on classic compute, 250 or fewer tables per pipeline is recommended, and a failover that loses the replication slot forces a full refresh of every table.
Open in REbricked →- Category
- Data engineering
- Also known as
- Postgres connector, PostgreSQL ingestion connector, Lakeflow Connect PostgreSQL
- Verified
- 2026-10-20