Operate on Databricks in ~5 minutes¶
What you'll do: stand up a small synthetic project on your own Databricks workspace, deploy it, run it, and query the result — driving the ade-ops engine end to end. No Fabric, no Power BI, no migration: this module is the purest look at what the engine does.
You'll need: a Databricks workspace (Community Edition works) and a personal access token. That's it.
Video walkthrough
A short screen-recorded walkthrough of this module is on the way — it will be embedded here. Until then, the steps below are the full path.
1. Set up the playground project¶
The reference distribution ships a self-contained, Databricks-only project that
generates its own synthetic data (pure Spark — no external dataset). Follow the
getting started bootstrap, then point onboarding
at the databricks-only scenario:
It routes you to distributions/reference/projects/playground/ — a synthetic
data generator plus one analytics notebook, designed to be operational in about
five minutes.
2. See what would change, then push¶
ade-ops never writes to a remote without showing you the diff first.
python -m core.cli status # env × scope overview
python -m core.cli push --env dev --scope notebooks --dry-run # preview
python -m core.cli push --env dev --scope notebooks # upload (after you confirm)
The --dry-run shows exactly what will land. The real push asks for explicit
confirmation — this human gate is the heart of the framework.
3. Run it¶
This seeds the synthetic tables and runs the analytics notebook on your workspace.
4. Query the result¶
You've now driven the full loop — author → preview → push → run → query — on your own Databricks, with every remote write gated. That loop is the same regardless of which BI layer (if any) you add later.
Where to go next¶
- Add a BI layer: serve this Databricks gold layer to Power BI — see the
databricks-to-powerbiquickstart. - Understand the machinery: the assembly pipeline.
- Go multi-environment: promote DEV → CERT → PROD with the human-gated push (module landing soon).