ATX AnalyticsTable gives analysts a local SQL client for Databricks SQL warehouses: Unity Catalog-aware connections, query tags for cost tracking, and a DuckDB copy of each result on the analyst's machine for pivots, charts and scheduled reports — with personal data masked before anything leaves it.
The Databricks connection form in ATX AnalyticsTable 1.7, filled in with the example values below. Secrets go into the encrypted vault, never into the project.
Unity Catalog grants for the service principal: -- The principal is its application ID; also give it CAN USE on the SQL warehouse. GRANT USE CATALOG ON CATALOG main TO `00000000-0000-0000-0000-000000000000`; GRANT USE SCHEMA ON SCHEMA main.finance TO `00000000-0000-0000-0000-000000000000`; GRANT SELECT ON SCHEMA main.finance TO `00000000-0000-0000-0000-000000000000`;
Reporting query — e-mails hashed in Databricks SQL: SELECT date_trunc('MONTH', t.booking_date) AS month, t.product_line, sha2(lower(c.email), 256) AS customer_key, sum(t.amount) AS amount FROM main.finance.transactions t JOIN main.finance.customers c ON c.customer_id = t.customer_id WHERE t.booking_date >= add_months(current_date(), -6) GROUP BY ALL ORDER BY month;
Then reshape locally — a Run SQL step on the project's DuckDB: -- "input" is the table the previous pipeline step produced (the import above). -- This runs on your machine, in the project's DuckDB, not on the database. SELECT region, SUM(revenue) AS revenue, COUNT(DISTINCT customer_key) AS customers FROM input GROUP BY region ORDER BY revenue DESC; — Column names follow the query above; on the local engine the pipeline's current table is registered as input before the script runs.
Unity Catalog column masks and row filters are evaluated for the principal that runs the query — the service principal above, for everyone. On Azure Databricks, an Entra ID mode with on-behalf-of runs each query as the signed-in person, so masks and filters apply per user. On AWS and Google Cloud workspaces that is not available.
Databricks runs in your cloud account, so queries need a route to the workspace. After the import, the data and all analysis over it are on your machine and work offline; the licence validates offline.
It started a stopped SQL warehouse. The socket timeout defaults to 15 minutes for that reason; serverless warehouses start in seconds.
Yes, for the principal that runs the query. With OAuth M2M that is the service principal for everyone; with Entra on-behalf-of on Azure, it is each signed-in person.
Yes, but it is a person's token and most organisations forbid it for services. OAuth machine-to-machine is the recommended method on every cloud.
This site needs JavaScript for the full experience. Every page above is reachable without it.