SQL workspace
The SQL workspace is a SQL tab in the data tables pane that lets you query the data behind your map with spatial SQL. It runs a full DuckDB engine, with its spatial extension, entirely in your browser - your query and your data never leave the page.
Querying your layers
Every data-bearing layer on the map is registered as a view named layer_... (based on the layer's name). Each view exposes the layer's columns plus a spatial geom column, so you can filter, join, aggregate, and run spatial functions across layers - for example, joining points to the polygons that contain them, or buffering and counting.
Write your query in the editor and press Ctrl+Enter (or the Run button) to run it. Results appear in the familiar data-table grid. For a bare SELECT, the preview is limited to the first 10,000 rows to keep the grid responsive.
Using the results
From a result set you can:
- Copy the rows as CSV or TSV.
- In editing mode, save to OneLake as GeoJSON and add the result to the map as a Shapes from file layer. The SQL that produced the layer is kept with it as provenance, so you can always see - and re-run - the query behind a derived layer.
Query history
Your recent queries are kept in a history list so you can re-run or refine them. History is stored per browser, on your machine - it isn't saved into the map item or shared with other authors.
Private by design
The SQL workspace is built for security-conscious environments:
- No query or data leaves the browser. All SQL executes client-side; there is no query service and no new network endpoint.
- The DuckDB engine and its spatial extension are served from the workload's own origin - nothing is fetched from a third-party CDN.
- Reads from OneLake use the same user-delegated access as the rest of Icon Map, so you can only query data you already have permission to see.
See the security whitepaper for the wider data-flow picture.
Availability
Like the rest of the data tables pane, the SQL workspace is an authoring aid: it's available in editing mode and is not shown in the published viewer.
Next steps
- Data tables - inspect the rows behind each layer.
- Shapes from file - the layer type created from saved results.
- Security whitepaper - where data does and doesn't flow.