pg_duckdb: Ducking awesome analytics in Postgres
summary
pg_duckdb embeds DuckDB inside Postgres so analytics run with DuckDB’s columnar engine, multithreading and Parquet/S3/Azure I/O instead of Postgres’ row scans. Three modes: run DuckDB on Postgres tables, read remote columnar files, or offload to MotherDuck. Benchmarks show big wins — ClickBench queries often ~2x faster and an anecdote: a TPC‑DS query on 10 GB drops from ~10 minutes to 450 ms vs 10 minutes. reads Parquet columns only for lower bandwidth.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 7:15What does pg_DuckDB look like?
- 8:08Differences between Postgres & DuckDB
- 13:10How does pg_DuckDB work?
- 15:193 ways to us pg_Duckdb
- 16:14DuckDB on Postgres tables
- 18:53Read from blob storage
- 22:06Copy data to MotherDuck
related talks
21mWhat can Postgres learn from DuckDB?
03mpg_duckdb: DuckDB-powered Postgres
24mpg_duckdb in Action: Accelerating Analytics on Azure Database for PostgreSQL
19mIn-Process Analytical Data Management with DuckDB
01h 02mAnalytics for not-so-big data with DuckDB
34mData Wrangling with DuckDB
27mIn Process Analytical Data Management with DuckDB
40mDuckDB: Harnessing in-process analytics for data science and beyond