Clarification on Synchronization of Databricks Unity Catalog Column Tags with Microsoft Purview Column-Level Tags

Hi Team, Any update on the above question?

Hi **SudhakarReddy Marepalli,
**Microsoft Purview does not synchronize Databricks Unity Catalog column‑level tags in real time. The integration between Unity Catalog and Microsoft Purview is scan‑based, not event‑driven. Any metadata updates made in Unity Catalog, including column‑level tags, are reflected in Purview only when a Purview scan is executed, either manually or through a scheduled run. This behavior is by design and applies to all metadata ingested through the Azure Databricks Unity Catalog connector.

To add or update column‑level tags in Unity Catalog, use the supported SQL syntax below:

ALTER TABLE <catalog>.<schema>.<table>
ALTER COLUMN <column_name>
SET TAGS ('<tag_name>' = '<value>');

After updating the tags in Unity Catalog

Trigger an Incremental Scan in Microsoft Purview. Navigate to

 Purview Studio → Data Map → Sources → Azure Databricks Unity Catalog source

Then either run the existing scan or edit the scan configuration to ensure it is set to Incremental scan. An incremental scan is supported and is the recommended approach for capturing metadata changes after the initial scan. A full scan is not required unless you need to re‑ingest all metadata from scratch.

Once the incremental scan completes, you can validate the result in Purview by opening the Databricks table asset, navigating to the Schema tab, selecting the column, and checking the Tags section. The Unity Catalog column‑level tags will appear there as technical metadata.

Unity Catalog column‑level tags are not automatically mapped to Microsoft Purview classifications or custom metadata attributes. Microsoft documentation does not define or guarantee any direct mapping between Unity Catalog tags and Purview governance constructs such as classifications. As a result, even when tags are ingested, they appear only as technical metadata fields and not as governed Purview classifications.

The primary reason for perceived tag synchronization issues is a misunderstanding of how the integration works. Many users expect tag updates in Unity Catalog to propagate automatically or immediately to Purview, but Purview updates its catalog strictly during scans and only for metadata fields that the connector explicitly supports. The connector also does not document overwrite or merge behavior for column‑level tags during incremental scans.

Proper permissions are required for metadata updates to propagate correctly. The identity used for scanning must have

USE CATALOG, USE SCHEMA, and SELECT permissions on the relevant tables and views. If data classification is enabled, SELECT permission is also required to allow Purview to sample data.

Purview connects to Databricks using a Databricks SQL Warehouse, which must be running at the time of connection and scan execution.

If consistent and deterministic synchronization of Unity Catalog column‑level tags into Purview classifications or custom attributes is required, the recommended best practice is to treat Unity Catalog as the system of record. Column‑level tags can be extracted from Databricks using the INFORMATION_SCHEMA.COLUMN_TAGS view and then programmatically published into Purview using APIs or SDKs. This is necessary because the native connector does not guarantee tag‑to‑classification mapping or ongoing synchronization behavior.
Reference Links:
Microsoft Purview – Connect to and manage Azure Databricks Unity Catalog

Databricks – Apply tags to Unity Catalog securable objects

Databricks – INFORMATION_SCHEMA.COLUMN_TAGS

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.

Please do not forget to "up-vote" wherever the information provided helps you, as this can be beneficial to other community members.

Hi Sai Jagadeesh,

Thank you for your previous clarification.

We would like confirmation regarding how metadata from Azure Databricks Unity Catalog (comments and tags) is synchronized into Microsoft Purview after a scan.

1️⃣ Comments (Table & Column Level)

• As per Microsoft documentation, our understanding is:
  • When table-level or column-level comments are updated in Unity Catalog and a Purview scan is executed
  • The updated comments should automatically populate into Purview under the corresponding table or column Description field.
  Could you please confirm:
  • Is this the expected behavior?
  • Does Purview overwrite the existing description if it was previously edited directly in Purview?
  • Is the Unity Catalog comment treated as the authoritative source during scan refresh?
• second question:
  • Suppose if I update column description in purview and at later time if I update the column comments in unity catalog, after scan runs, will it overwrite the existing description in purview or keep the existing description.

  2️⃣ Tags in Unity Catalog

  We also need clarification on how Unity Catalog tags are reflected in Purview. When tags are applied at table or column level in Unity Catalog:
  • Do these tags automatically appear under the Column Tags section in Purview?
  • Or Are they stored as Custom Metadata (Data Asset Attributes)?
  • Or Are they ingested as Purview classifications?
  • Is there any mapping configuration required for tag synchronization?

Thanks,

Sudhakar

@SAI JAGADEESH KUDIPUDI Any update on this please.

Hi SudhakarReddy Marepalli,
Q: When table‑level or column‑level comments are updated in Databricks Unity Catalog and a Microsoft Purview scan is executed, should the updated comments automatically populate in Purview?

A: Yes. This is the expected behavior. When Microsoft Purview scans Azure Databricks Unity Catalog, it ingests technical metadata from Unity Catalog. As part of this process, Unity Catalog metadata is mapped as follows:

• Unity Catalog table comments are ingested into the Asset Description field in Purview (table level).
• Unity Catalog column comments are ingested into the Column Description field in Purview (column level).

During each scan (full or incremental), Purview refreshes this metadata from Unity Catalog and updates the corresponding description fields accordingly.
Q: Does Purview overwrite an existing description if it was previously edited directly in Purview?

A: Yes. If a description (table or column) was manually edited directly in Microsoft Purview, it will be overwritten during a subsequent scan if Unity Catalog contains a value for the corresponding comment.

Purview treats Unity Catalog as the authoritative source for scanned metadata fields. During scan refresh, Unity Catalog values take precedence and replace any manually maintained descriptions in Purview for those same fields.
Q: Is Unity Catalog treated as the authoritative source during scan refresh?

A: Yes. For Azure Databricks Unity Catalog sources, metadata synchronization is one‑way from Unity Catalog to Microsoft Purview. Unity Catalog is considered the system of record for scanned metadata such as table and column comments. Updates made in Purview do not flow back to Unity Catalog.
Q: Suppose if I update column description in purview and at later time if I update the column comments in unity catalog, after scan runs, will it overwrite the existing description in purview or keep the existing description.
After the next Microsoft Purview scan runs, the existing column description in Purview will be overwritten with the updated column comment from Databricks Unity Catalog.

This is because the integration is scan‑based and one‑way, and Unity Catalog is treated as the authoritative source for table and column comments during Purview scans. Any manual edits made directly in Purview for those description fields are replaced when the scan refreshes metadata from Unity Catalog.
Q: When tags are applied at the table or column level in Databricks Unity Catalog, how are they reflected in Microsoft Purview?

A:

• Unity Catalog tags automatically appear in Microsoft Purview as Tags under the Schema → Column Tags (for column‑level tags) and Table Tags (for table‑level tags) after a Purview scan is executed. They are synchronized during the next full or incremental scan.
• Unity Catalog tags are not ingested as Purview classifications. Purview classifications are a separate feature and are applied either automatically by Purview’s classification rules or manually. Unity Catalog tags do not map to built‑in or custom classifications.

• Unity Catalog tags are not stored as Custom Metadata (Data Asset Attributes) in Purview. They are displayed specifically in the Tags section of the asset and column metadata views.

• No additional mapping or configuration is required for tag synchronization. The integration is scan‑based and one‑way. As long as the Azure Databricks Unity Catalog source is registered and scanned in Purview, tags are automatically ingested during scan execution.

The provided Q&A accurately reflects Microsoft Purview's integration with Azure Databricks Unity Catalog as of early 2026 documentation

Reference Links:
Connect to and manage Azure Databricks Unity Catalog in Microsoft Purview