Presented by:

Ce6b151bd259224b8269e9be1db9bc1c

Zhou Sun

Mooncake Labs
No video of the event yet, sorry!

In this talk, we'll explore the various attempts to make Postgres more effective for analytics.

Early projects like Citus and Timescale added columnar storage, but lacked a strong vectorized execution engine to fully leverage it.More recent projects embed DuckDB as an execution engine within Postgres. pg_mooncake is one such attempt that introduces a columnstore table with DuckDB execution.

We'll also share insights on how we built the pg_mooncake extension and get to top 10 of clickbench (benchmark for all analytics systems).

Date:
Duration:
20 min
Room:
Conference:
Postgres Conference 2025
Language:
Track:
Variants and Cloud
Difficulty:
Medium