Updated
What happened to ScratchDB?
If you're looking for ScratchDB, the open-source data warehouse on ClickHouse from 2023: the project now lives on GitHub as ScratchData, and this domain is home to a different site, a directory of developer and AI-builder tools. This page explains both and lists the closest alternatives today.
What the original ScratchDB was
ScratchDB launched on Hacker News in October 2023 as “Open-Source Snowflake on ClickHouse”. It was a managed data warehouse with two HTTP endpoints: one to insert JSON and one to query with SQL. It created tables and columns automatically from the shape of the JSON, batched the data and bulk-loaded it into ClickHouse.
It was written in Go, released under the AGPL-3.0 license, and built by Scratch Data, a Y Combinator company.
Where the project is now
The code is still open source at github.com/scratchdata/ScratchDB, now called ScratchData. It has grown from a single ClickHouse warehouse into a tool that streams JSON into several analytics databases.
ScratchDB.com is an independent tools directory and isn't affiliated with Scratch Data or the ScratchData project.
Alternatives to ScratchDB today
- ScratchData (the original project)
The same team's open-source Go project, now a wrapper that streams JSON into DuckDB, BigQuery, Snowflake, Redshift, ClickHouse or MotherDuck and creates tables and columns automatically.
- ClickHouse
The open-source column-oriented database ScratchDB was built on. Run it yourself, or use ClickHouse Cloud for a managed service.
- Tinybird
A managed platform on ClickHouse with HTTP ingestion and SQL-to-API endpoints, the closest hosted match to ScratchDB's “send JSON, query with SQL” model.
- DuckDB
An in-process analytical database. A good fit when your data fits on one machine and you don't want to run a server.
- MotherDuck
A hosted, serverless service built on DuckDB, for sharing analytical data and queries across a team.
What ScratchDB.com is now
A directory of databases, auth, hosting, payments, UI kits and AI tools for people who build software, with side-by-side comparisons.