Deploying a bundle through REST APIs

Databricks Asset Bundles are deployed with the databricks bundle deploy command, which is a thin client that translates your bundle's YAML configuration into a series of calls against standard Databricks REST APIs (Workspace, Jobs, Pipelines, etc.). There is no separate 'Bundle' REST endpoint—CLI/CI tooling always sits on top of the same public REST APIs used elsewhere in the platform.

1 · Learn the must-know

• 1databricks bundle deploy authenticates using the same mechanisms as any REST API call—PAT, OAuth token, or a service principal—typically supplied via DATABRICKS_HOST/DATABRICKS_TOKEN env vars or a configured CLI profile.
• 2During deploy, source files and notebooks are uploaded to the workspace via the Workspace API, while jobs, pipelines, and other resources are created or updated via their respective REST APIs (Jobs API, Pipelines API, etc.).
• 3Deployment is idempotent: bundle state tracks previously deployed resource IDs so re-running deploy updates existing jobs/pipelines instead of creating duplicates.
• 4Which workspace/host the REST calls target is controlled by the targets section in databricks.yml (e.g., dev, staging, prod), so the same bundle can deploy to different environments without code changes.
• 5In CI/CD pipelines you typically still invoke the Databricks CLI to perform bundle deploy rather than hand-crafting raw REST calls, since the CLI manages the correct sequencing and dependency resolution of the underlying API operations.
• 6Permissions for the deployed resources (jobs, pipelines) are set via the REST APIs' permission endpoints as declared in the bundle's permissions block, so the deploying identity must have sufficient rights in the target workspace.

2 · Check your understanding