Tutorial: Mirrored Databricks catalog editor¶
CSA Loom
mirrored-databrickseditor — mount a Databricks Unity Catalog as a read-only mirror into ADLS Gen2 Delta and query it via Synapse. No Microsoft Fabric or OneLake required.
What it is¶
A Mirrored Databricks catalog brings a Databricks Unity Catalog into Loom analytics as a read-only mirror. The UC Delta tables are mirrored into ADLS Bronze (ADF CDC / Synapse Link) and queried via the Synapse serverless SQL analytics endpoint, without re-ingesting or copying governed data. Fabric mirroring into OneLake exists only as an explicit opt-in — never the default.
When to use it¶
- Databricks owns the governed data (Unity Catalog) and you need it queryable alongside Loom lakehouses and warehouses.
- You want cross-engine joins (Synapse SQL over UC Delta) without an ETL copy job to maintain.
Step-by-step in Loom¶
- Create the item. Choose + New item → Mirrored Databricks catalog (Data Factory). The editor opens at
/items/mirrored-databricks/<id>. - Provide the workspace. Point at the Azure Databricks workspace and Unity Catalog you want to mirror.
- Select the catalog/schema. Choose which catalog and schemas to expose as a read-only mirror in ADLS Bronze Delta.
- Query via Synapse SQL. Read the mirrored Delta tables via the Synapse serverless SQL analytics endpoint or Spark — no copy required.
- Respect source governance. Mirroring is read-only; writes and permissions stay governed by Unity Catalog on the Databricks side.
The Azure backend it rides on¶
- Mirror: ADF CDC / Synapse Link copy into ADLS Gen2 Bronze Delta.
- Query: Synapse serverless SQL (analytics endpoint) or Spark over the mirrored Delta.
- Governance: Unity Catalog remains the source of truth on the Databricks side.
No Fabric required¶
The default mirror path is ADLS + Synapse; Fabric/OneLake mirroring is opt-in only.
Learn more¶
- Databricks mirroring (parity source): https://learn.microsoft.com/fabric/database/mirrored-database/azure-databricks-tutorial