Skip to main content

DABs

You'll learn how to deploy a full medallion pipeline using Databricks Asset Bundles in ~15 min.

Prereqs: Build the first pipeline, DABs CLI

Why this matters​

The UI is great for prototyping, but production pipelines need to be version-controlled, testable, and deployable across environments. Databricks Asset Bundles (DABs) let you define pipelines, jobs, schemas, and compute as code in a single project. One databricks bundle deploy command provisions everything.

How it works​

The example project below deploys a complete medallion architecture pipeline, including:

  • 1 job with 2 tasks: a notebook task that generates sample data, followed by a pipeline task that runs the transformation.
  • 1 Spark Declarative Pipeline — the bronze-silver-gold transformation chain.
  • 3 schemas — bronze, silver, and gold.
  • 1 volume — a landing zone for raw files.
  • 1 DBSQL warehouse — for querying the output.

See all DABs supported resources.

Medallion Architecture on Unity Catalog using DABs​

The GitHub repository is ready to clone and deploy:

github.com/ivancalvo-dbxs/medallion-pipeline-dabs

Video walkthrough​

When to use / when not to​

Use DABs when:

  • You want version-controlled, repeatable deployments.
  • The pipeline needs to run in dev, staging, and production.
  • You want to define all resources (schemas, volumes, warehouses) alongside the pipeline code.

Use the UI + agent instead when:

  • You are exploring data and prototyping transformations.
  • You want fast iteration without setting up a local development environment.

Next​