pg_duckdb in Action: Accelerating Analytics on Azure Database for PostgreSQL | POSETTE 2026

Nitin Jadhav walks through how pg_duckdb enables analytics directly inside PostgreSQL, and demonstrates it running on Azure Database for PostgreSQL.

Overview

The talk focuses on using pg_duckdb to accelerate analytical workloads in PostgreSQL without standing up a separate data warehouse or building ETL pipelines just to analyze data that already lives in Postgres.

Key idea: instead of exporting data out of Postgres for analytics, pg_duckdb lets teams run fast, columnar-style analytical queries inside Postgres, and also query Parquet data from object storage without first loading it into Postgres.

What pg_duckdb is and where it fits

PostgreSQL workload strengths

The presenter frames PostgreSQL as strong for transactional workloads (OLTP), then contrasts that with analytical workloads (OLAP) that often require different execution characteristics.

OLTP vs OLAP

Why analytics can be hard in Postgres

The talk highlights that teams often end up building ETL pipelines and moving data out of Postgres to answer analytical questions, due to limitations in how Postgres typically handles analytics-heavy query patterns.

DuckDB overview and why it helps

The presenter introduces DuckDB and its advantages for analytics, positioning it as a way to get columnar-style performance characteristics for analytical queries.

How PostgreSQL processes queries (high level)

A segment of the talk explains PostgreSQL query processing at a conceptual level to set context for where acceleration can be gained when running analytics-style queries.

pg_duckdb capabilities (including Azure scenarios)

The talk covers pg_duckdb capabilities such as:

Demo: pg_duckdb on Azure Database for PostgreSQL

The demo showcases:

Practical takeaways