Why it matters
A new lakehouse, or a move of an existing one, is usually designed in documents and tested in production. Problems surface after the pipelines are live. Lakehouse Studio lets you design every layer, test it end to end on generated data, and hand over reviewed code, before anything runs on your platform.
What you can do
- Start a project for a new lakehouse, an existing LakeLogic lakehouse, or a platform you want to bring into LakeLogic.
- Group sources into domains and design bronze, silver and gold.
- Test the whole design on synthetic rows. No production data is read.
- Explore contracts as tables, and preview a contract change, on synthetic data.
- Deliver the contracts and a pipeline project as a pull request.
How to do it in LakeLogic
- Go to Build → Lakehouse Studio → Projects and select Start new project.
- Choose a starting point: Build a new lakehouse, Extend or update a LakeLogic lakehouse or Bring an existing platform into LakeLogic. Optionally pick a Target platform: Databricks, Microsoft Fabric, Snowflake or Contracts only. You can choose Decide later.
- Select sources. The project reads your sources or landing zone. Only metadata is observed.
- Design & test. Work through the design steps:
- Domains: accept how sources group into domains.
- Bronze: raw entities mirrored as strings.
- Silver: types, quality rules and joins.
- Gold: facts and dimensions.
- Lineage, then Test: a test run of the whole design on synthetic rows. Resolve open questions before a layer is approved.
- Prepare delivery. Review the design and emit the contracts and a pipeline project as a pull request in your repository.
To try things without a project, open Explore & test. It shows your contracts as tables built from synthetic data, so you can test changes before they ship.
What you get
- One contract file per asset, plus a pipeline project, in a pull request.
- A test result for the whole design: passed, or the number of assets that failed.
- A project record of its sources, design revisions and delivery.
Good to know
- Delivery ends at a pull request. LakeLogic does not run or deploy the code. You merge it, and your own CI runs it on your platform, bronze to silver to gold in dependency order.
- Synthetic means shape, not truth. A passing test shows the design runs end to end. It does not show your real data will pass the rules.
- Explore & test appears only if Lakehouse Studio is enabled for your organisation.
- Drafted contracts stay in the project until their pull request merges. They do not appear in Contract Studio before then.
- To start from what you already run, first know what data you have.