Microsoft Purview Data Governance is Microsoft’s data governance solution, built on the Data Map and Unified Catalog. Microsoft is retiring it. In Microsoft’s November 12, 2026 announcement, Microsoft said support ends November 12, 2027, and governance moves to Fabric.
Microsoft Learn adds that new accounts are closed and existing features stay in maintenance mode until then (Data Map data sources, updated 2026-11-12). Until that date, the Data Map’s own grid shows what exists.
If you run governance on Purview, you face a clear job. List every source you scan, and mark which capabilities you rely on. Skip it, and you migrate without knowing what must be covered.
The retirement guide has the dates and decisions. The migration plan covers the move. The rest maps what the grid lists, source by source, and what replaces it.
Microsoft’s grid tracks five things per source: scan, classify, label, apply policy and trace lineage. Search results often fold them into one yes or no. The grid keeps them apart.
Four points to know before you check your own stack:
- Scan, classify, label, policy and lineage are five separate columns, not one score.
- The grid lists policies for Azure sources and for SQL Server on Azure Arc. That one is Microsoft’s own database engine, managed through Arc.
- Lineage listed in the grid often means lineage inside that one system.
- The File category lists Amazon S3 and HDFS. Azure Files sits under the Azure sources.
| Topic | What the grid and connector pages list |
|---|---|
| What it is | Data Map registering more than 20 non-Microsoft sources for scan, classification, labels, policy and lineage |
| Policies | Azure sources and SQL Server on Azure Arc |
| Lineage depth | Within Snowflake; notebook-based for Azure Databricks Unity Catalog; metadata scan for Amazon Redshift |
| File shares | Amazon S3 and HDFS in the File category; the Information Protection scanner reaches on-premises shares for labeling |
| Recent movement | DLP for non-Microsoft SaaS apps reaches general availability in late December 2026, moved from late October |
| Best for | Estates broader than Microsoft alone; check each source’s five capabilities, then plan what replaces the Data Map |
What is Microsoft Purview, and what is data governance?
Microsoft Purview is a set of data security and compliance tools. Data governance ensures the data you use is “discoverable, accurate, trusted, and protected”, per Microsoft’s overview (data governance overview).
Where can I find Microsoft’s data governance solutions?
Today they are the Data Map and Unified Catalog in the Purview portal. Going forward, Microsoft points governance to the OneLake catalog in Fabric.
What does Microsoft Purview’s Data Map actually cover?
The Data Map tracks five capabilities for each registered source. Per Microsoft’s data sources page, more than 20 non-Microsoft sources are registered.
Snowflake, Azure Databricks Unity Catalog, Google BigQuery and Amazon Redshift are the four most teams ask about. Each gets its own section below.
What data does Microsoft Purview protect?
Purview’s security side protects data with sensitivity labels and data loss prevention. The Data Map and Unified Catalog hold metadata about data, not the data itself, per Microsoft’s overview.
What types of capability does the Data Map track?
Each capability answers a different question:
- Scan: can the source be registered and its metadata read?
- Classify: does the grid list automatic classification?
- Label: can sensitivity labels attach to Data Map assets?
- Policy: does the grid list policies for the source?
- Lineage: what lineage does the connector capture?
A source showing up in the Data Map answers “is it registered”. It does not answer “is it governed like a native Azure resource”. The grid answers that second question, one column at a time. For the baseline, read what Microsoft Purview is.
How do the Data Map, Unified Catalog and agent controls fit together?
The Data Map scans and registers. Purview Unified Catalog curates what the Data Map finds into governance domains and data products. Purview controls for AI agents secure what an agent can do with Fabric data. The OneLake catalog is Fabric’s own discovery layer.
The same line separates any catalog from a context layer, as in data catalog versus context layer. It matters more once an AI agent reads the result. See the types of metadata for AI agents.
How does Purview work with non-Microsoft sources like Snowflake and Databricks?
The Data Map scans and registers them, and the depth varies by capability. The table shows what Microsoft lists for each source (Data Map data sources, ms.date 2026-10-13).
| Source | Classification | Sensitivity labels | Policies | Lineage |
|---|---|---|---|---|
| Snowflake | Automatic | Listed | Azure sources only | Within Snowflake, plus pipeline hops |
| Azure Databricks Unity Catalog | Automatic (custom classification listed separately) | Not listed | Azure sources only | From notebook runs |
| Google BigQuery | Not listed | Not listed | Azure sources only | Within BigQuery |
| Amazon Redshift | Not listed | Not listed | Azure sources only | Not listed |
“Azure sources only” means the grid lists policies for Azure sources and SQL Server on Azure Arc. “Not listed” means the grid’s column is blank for that source. Registered and governed to Azure depth are different claims.
Does Microsoft Purview work outside Azure?
Yes for registration and scanning. Capability differs by source, and policies appear for Azure sources and SQL Server on Azure Arc.
For Snowflake or Databricks, the grid lists scanning, classification and lineage options. Access control for those platforms stays in their own permission models. How access control works in an AI context platform covers one approach. Who governs a Fabric item an AI agent created covers a related question.
How deep is Purview’s lineage across Snowflake, Databricks, BigQuery and Redshift?
“Lineage” in the grid covers four different behaviors. Check which one applies to your source.
Snowflake: lineage inside Snowflake
Microsoft’s Snowflake connector page describes “static lineage on asset relationships among tables, views, streams, and stored procedures.” That is lineage between Snowflake objects.
Lineage that crosses systems needs the dataset to act as a source or sink in an Azure Data Factory or Synapse pipeline. Snowflake to Fabric migration covers the same pipeline-hop rule.
Azure Databricks Unity Catalog: lineage from notebooks
Per Microsoft’s Unity Catalog connector page, Purview captures lineage for tables, views and columns connected through Databricks notebooks. It reads the system.access.table_lineage and system.access.column_lineage system tables, so they must be enabled first.
Column-level lineage appears when a notebook runs on a cluster. SQL warehouses do not generate it, a point Databricks’ lineage documentation confirms.
Google BigQuery: a scan scoped by region
Microsoft’s BigQuery connector page says it scans datasets in the US multi-regional location. A dataset in us-east1 or an EU region completes the scan with zero assets. Check the page before you rely on the scan outside the US.
Amazon Redshift: a metadata scan
The grid lists Redshift for metadata scanning. Its classification, label, policy and lineage columns are blank. “Purview registers Redshift” and “Purview governs Redshift” are different sentences.
Match lineage depth to your needs. Column-level lineage and a lineage proof of value are the same discipline on the Fabric side. The Databricks view of the question is in Genie ontology and the Atlan context layer, Databricks Omnigent versus a neutral context layer and Unity Catalog versus a cross-platform semantic layer. Synapse to Fabric migration shows the pipeline hop again.
Can Microsoft Purview scan on-premises file shares or network drives?
The Data Map’s File category lists two sources: Amazon S3 and HDFS. Azure Files sits under the Azure sources, because Microsoft runs it.
A Microsoft Q&A thread from April 2025 shows the usual question. A named asker, Fabio Santos, asked whether the Self-hosted Integration Runtime can scan on-premises file servers. A reply named the options: use the Information Protection scanner, or sync files to Azure storage and scan there. These are community replies, not official documentation.
How does the Information Protection scanner differ from the Data Map?
The scanner is a separate product. Per Microsoft’s scanner documentation, it runs on Windows Server and finds, classifies and protects files on network shares and SharePoint Server libraries. It labels files for data loss prevention. Registering assets and cataloging schema are Data Map jobs.
| Capability | Data Map scan | Information Protection scanner |
|---|---|---|
| Registers the asset in the Data Map | Azure Files, Amazon S3, HDFS | Not its job |
| Extracts and catalogs schema | Azure Files, Amazon S3, HDFS | Not its job |
| Classifies and labels files for DLP | Classification where listed | Yes |
| Reaches on-premises SMB or NFS shares and SharePoint Server | Not its job | Yes |
A team that sees the scanner running on a shared drive may assume the drive is in the Data Map. They are two products. What lives on an ungoverned shared drive is covered in tribal knowledge. Fabric governance shows the parallel “built in versus add” split.
How do you set up Purview for Fabric, and how do I get started?
Governance setup now happens in Fabric, and Purview setup is about protection.
Microsoft’s Fabric guidance says the OneLake catalog is “the primary Fabric experience for discovering and governing Fabric data” (Use Microsoft Purview with Fabric, ms.date 2026-08-26). Using Purview “doesn’t change where Fabric governance lives.”
Purview Data Governance is retiring and new accounts are closed, so there is no new Data Map to start. What exists today:
- Governance in the OneLake catalog. Discovery, insights, recommended actions, domains, endorsement and lineage are built in. Microsoft says governance is included in Fabric capacity. See the OneLake catalog guide and Fabric governance.
- Sensitivity labels. Purview Information Protection sets labels on Fabric items.
- Data loss prevention policies. They cover lakehouses, warehouses, databases and semantic models.
- Audit logging. Fabric user activity appears in the Purview audit log.
- Insider risk management. Policies include ready-made indicators for Fabric activity.
- An existing Data Map scan of Fabric. The grid lists Fabric for lineage and live view. Existing features stay in maintenance mode through November 12, 2027. Plan the move with the migration plan.
What are the limits of Purview’s coverage, and what is changing?
The limits below are the edge of the Data Map until support ends on November 12, 2027. Microsoft Learn says the governance features are in maintenance mode, so the gaps are not on a closing path.
- Policies: listed for Azure sources and SQL Server on Azure Arc.
- Lineage: often inside one system; cross-system lineage uses Data Factory or Synapse pipelines.
- File shares: Amazon S3 and HDFS in the File category.
- Region: the BigQuery scan covers the US multi-regional location.
Microsoft’s direction is the OneLake catalog. The announcement says it governs data across Fabric, Azure and third-party sources through shortcuts, mirroring and supported connectors, without copying data.
On the security side, DLP and auto-labeling for non-Microsoft SaaS apps is still rolling out. Per the message center post, updated October 30, 2026, DLP covers Box, Dropbox, Google Workspace, Salesforce and Cisco Webex. Auto-labeling covers Google Workspace and Box. General availability moved from late October to late December 2026. This is a labeling and DLP expansion. It does not add Data Map sources.
What’s new in Microsoft Purview tracks changes. Re-check the message center before you trust any date, because this one has moved.
How do you check whether Purview’s coverage is enough for your stack?
Treat this as a self-check against your own source list. With the Data Map retiring, it also becomes your requirements list for the move.
Examples: reading the grid for your own sources
- Is the source in the Data Map’s list? Check Microsoft’s current grid.
- Which of the five capabilities does the source need? Scan alone may suit a low-risk source. A regulated source may need policies.
- Is your lineage within one system or across systems? Cross-system lineage runs through a Data Factory or Synapse pipeline.
- Where does unstructured data sit? Azure Files, Amazon S3 and HDFS are in the Data Map. Other shares use the scanner.
- What does “covered” cost? Only assets linked to a governance concept are governed and billed. In Microsoft’s billing example, a SQL Server with 200 tables and 20 linked tables bills 20 governed assets.
- Does the SaaS DLP expansion reach your content stores? Check the message center for the current date.
Run this list against your sources, not against Purview’s category page. The context layer evaluation criteria use the same discipline. A single platform’s native coverage may not stretch across an estate, the situation in single-stack lock-in versus a neutral context layer. For the multi-cloud version, see governing AI agents across multiple clouds and the multi-cloud context layer.
For options, see Purview alternatives, best data catalogs for Microsoft Fabric and the Fabric governance evaluation questions. Pick the replacement from your own list.
What replaces the Data Map, and where does Atlan fit?
The Data Map list becomes the spec for whatever replaces it.
Atlan is an AI Context Platform that connects to more than 100 sources. Its Fabric connector is generally available and catalogs these items down to column level, with end-to-end lineage:
- Workspaces, lakehouses and warehouses
- Semantic models, reports and dashboards
- Dataflows and data pipelines
Atlan’s Context Lakehouse keeps context as open Apache Iceberg tables in OneLake. Fabric can query them directly. That context reaches people in Microsoft Teams and agents in Copilot Studio. Microsoft’s announcement names Atlan among the third-party providers Fabric works with, and Atlan is available in Microsoft Marketplace.
For the move, see the migration plan and the Atlan and Purview page. For the design, see the context layer reference architecture, the enterprise context layer and context engineering.
FAQs about Microsoft Purview’s data governance coverage
1. What does Microsoft Purview actually govern?
Until support ends on Nov 12, 2027, Purview Data Governance tracks five capabilities per source: scanning, classification, labeling, policy and lineage. A source can show up in some and not others, so check all five.
2. Does Microsoft Purview work with non-Microsoft data sources like Snowflake or Databricks?
Yes for registration and scanning. Classification and lineage vary by source. Microsoft’s grid lists policies for Azure sources and SQL Server on Azure Arc. This applies until support ends Nov 12, 2027.
3. Can Microsoft Purview scan on-premises file shares or network drives?
The Data Map’s File category lists Amazon S3 and HDFS. The Information Protection scanner reaches on-premises shares for labeling and DLP, per a Microsoft Q&A community reply. Registering shares in the Data Map is a Data Map job.
4. Is Microsoft Purview only for Azure and Microsoft 365?
No. More than 20 non-Microsoft sources register in the Data Map. Policies appear for Azure sources and SQL Server on Azure Arc. The Data Map retires, and Microsoft says the OneLake catalog governs Fabric, Azure and third-party sources.
5. Does Purview provide lineage outside of Microsoft Fabric and Power BI?
Yes, mostly inside a single system. Snowflake and Google BigQuery lineage is generated inside that platform. Cross-system lineage depends on a Data Factory or Synapse pipeline.
6. What are the real limitations of Microsoft Purview for a multi-cloud estate?
Three scope points: policies are listed for Azure sources and SQL Server on Azure Arc, lineage is usually within one system, and file shares use the scanner. These hold until support ends Nov 12, 2027, after which Microsoft points governance to Fabric.
7. Can Purview enforce access policies on Snowflake or Databricks?
Microsoft’s grid lists policies for Azure sources and SQL Server on Azure Arc, the latter being Microsoft’s own engine. For Snowflake and Databricks Unity Catalog it lists scanning, classification and lineage options instead.
8. What does the Purview Information Protection scanner actually do, and how is it different from the Data Map?
It finds, classifies and labels files on network shares and SharePoint Server for data loss prevention. Registering assets in the Data Map and cataloging schema are Data Map jobs.
9. Is Microsoft Purview enough on its own if we also run Databricks, Snowflake, or BigQuery?
The Data Map is retiring, so the question is what covers you after Nov 12, 2027. For scanning and some classification, the grid lists support today. For cross-platform policy and lineage across platforms, plan a layer beyond the Data Map.
10. How do you set up Microsoft Purview for Fabric?
Governance setup now happens in Fabric, where Microsoft says the OneLake catalog is the primary experience. Purview adds labels, DLP, audit and insider risk for Fabric. New Purview Data Governance accounts are closed.
11. Is Microsoft Purview Data Governance being retired?
Yes. Microsoft said in its November 12, 2026 announcement that support ends Nov 12, 2027. New accounts are closed, and Purview stays focused on data security and compliance.
Sources
- Microsoft’s November 12, 2026 announcement: a next step in the evolution of the Microsoft data governance strategy
- Data sources that connect to Microsoft Purview Data Map, Microsoft Learn (ms.date 2026-10-13, updated 2026-11-12)
- Connect to and manage Snowflake in Microsoft Purview, Microsoft Learn
- Connect to and manage Azure Databricks Unity Catalog in Microsoft Purview, Microsoft Learn
- Connect to and manage Google BigQuery projects, Microsoft Learn
- Purview Scan On-Prem File Server, Microsoft Q&A community thread
- Learn about the Microsoft Purview Information Protection scanner, Microsoft Learn
- Microsoft Purview: Data Loss Prevention and auto-labeling support for non-Microsoft applications (MC1449180), Microsoft 365 message center (general availability moved from late October to late December 2026, per the October 30, 2026 update)
- What’s new in Microsoft Purview, Microsoft Learn release notes
- Data governance overview, Microsoft Learn
- Use Microsoft Purview with Microsoft Fabric, Microsoft Learn
- Billing in Microsoft Purview Data Governance, Microsoft Learn
- Unity Catalog data lineage, Azure Databricks documentation, Microsoft Learn