A duck for your dashboard: Performant data apps in the browser with DuckDB
summary
Observable uses DuckDB in-browser and at build time to ship static data apps: data loaders (shell/Python/R) run offline to materialize parquet tables and an initial.json bootstrap, clients download the package and instantiate DuckDB in the browser. A PoC targets hundreds of customers with per-customer datasets in the millions to tens of millions of rows, so interactive queries require prepared query debouncing and a small rehydration file; wishlist items include faster modular client loading and native query-result materialization for quick render.
AI-written summary · about
chapters (7)
- 0:00DuckDB in Observable
- 3:52Observable Framework static
- 9:15Interactive dashboards with DuckDB
- 12:03Customer proof with DuckDB
- 15:57Wish list for DuckDB
- 18:25DuckDB use cases & conclusions
- 19:18Q&A
related talks
16mTaking Flight with Interactive Analytics
40mDuckDB Meets Data Lakes
29m"Serpents and Ducks: wrangling data with Python and DuckDB"
26mLife Beyond Pandas: Workflows with DuckDB, Daft, Polars, and Datafusion
45mUnderstanding geospatial data with duckdb
43mUse dbt and DuckDB Instead of Databricks for Data Processing
32mDash: Interactive Data Visualization Web Apps with no Javascript
36mEmbrace the chaos: Composable data systems with fewer asterisks