Skip to content

MotherDuck

Engine: DuckDB · Format: DuckLake · Storage: MotherDuck-hosted catalog · Status: ✅ Live.

MotherDuck is a serverless analytics platform built on DuckDB — more than hosted DuckDB. It adds hybrid execution (splitting a query between your laptop and the cloud), a managed cloud catalog with zero-copy sharing, a collaborative SQL UI, and elastic scale — all DuckDB-native. With DuckLake, your governed medallion runs from a laptop straight into a MotherDuck-hosted lakehouse — same contracts, no local files, no Spark.

Reference data mesh lakehouse: lakelogic-motherduck-data-mesh-lakehouse (coming soon — repo is private for now) — a deliberately thin, upload-and-run artifact: how to seed and build the mesh into MotherDuck, and every table it materializes.

The same contract

Identical to the DuckLake provider; only the catalog location changes (local file → MotherDuck):

materialization:
  strategy: merge
  format: ducklake        # metadata target points at MotherDuck ('md:')

Special configuration

How the LakeLogic framework adapts to this backend (handled for you):

  • Native DuckLake database, not ATTACH. MotherDuck "workspace mode" rejects ATTACH 'ducklake:md:…'. The framework instead connects to md: and issues CREATE DATABASE IF NOT EXISTS "<catalog>" (TYPE DUCKLAKE), then writes into it. The reference framework detects the MotherDuck backend automatically from an md: metadata path.