Notebooks
Most notebooks run Python, so the shortest path is the duckdb package inside the notebook: it reads your bucket objects, Iceberg tables and DuckLake tables in place, with no server to set up. Use a warehouse connection instead when your team already queries Send2Bucket data through one, or when a query is too large for the notebook's memory.
| Tool | Connect through | Page |
|---|---|---|
| Jupyter, JupyterLab, Google Colab, VS Code | DuckDB or PyIceberg in a Python cell; BigQuery magics in Colab | Jupyter |
| Hex | DuckDB in a Python cell; MotherDuck, Trino, Athena, BigQuery or PostgreSQL connections | Hex |
| marimo | SQL cells on DuckDB | marimo |
| Deepnote | DuckDB in a Python block; BigQuery, Athena, Trino, ClickHouse, PostgreSQL and other integrations | Deepnote |
| Observable | Framework data loaders that write Parquet with DuckDB; DuckDB database connector in Notebooks 2.0 | Observable |