Why it matters
You cannot govern what you cannot see. One connection and one Discovery run give you a list of your tables, which already have a contract, and draft contracts for the ones that do not.
What you can do
Connect a data platform to scan its catalog and read metadata:
| Platform | What LakeLogic reads |
|---|---|
| Databricks | Unity Catalog, and table and column lineage |
| Microsoft Fabric | Lakehouse tables and metadata |
| Snowflake | Databases, schemas and metadata |
| Google BigQuery | Datasets, tables and metadata |
| AWS Glue Data Catalog | Databases and tables behind Athena and Redshift Spectrum |
| Azure Synapse Analytics | Dedicated and serverless SQL pools and metadata |
You can also connect cloud storage (AWS S3, Azure ADLS Gen2, Google Cloud Storage) to check your landing zone exists and inspect raw files. Git, BI, messaging and ticketing tools are covered in Connect your tools.
How to do it in LakeLogic
You need an owner or admin role, and a workspace on a trial or paid plan (How a trial starts).
Connect
- Go to Admin → Integrations. This is the one place to create, change or remove a connection.
- Find your platform under Data Platforms or Cloud Storage and open its card.
- Fill in the fields. Each card links to where you find the values. Secrets are stored encrypted.
- Test the connection to confirm LakeLogic can reach the platform.
Run Discovery
- Go to Understand → Estate and open the Discovery tab.
- Choose what to scan. You can narrow it to tables matching a name pattern.
- Select Run discovery.
Discovery runs against Databricks, Microsoft Fabric, Snowflake and connected cloud storage.
What you get
- Your tables, reconciled against the contracts you already have.
- Draft contracts for tables that have none. They stay drafts until someone reviews them.
- A record of which platform each run scanned.
- Findings to prioritise on the Assessment tab, next to Discovery.
From here, see Know your estate and Write a data contract.
Good to know
- The Connections tab on Estate only shows what the estate can reach. Changes happen in Integrations.
- Hive Metastore is not offered. On EMR, connect AWS Glue. On Databricks, use the Databricks connector. On premises, run LakeLogic Core inside your network.
- A read-only workspace cannot add connections or run Discovery.