Skip to main content

Purview vs Unity Catalog: Comparison and Evaluation Guide

Emily Winks, Data Governance Expert, Atlan
Data Governance Expert
Updated:
|
Published:
18 min read

Key takeaways

  • Purview Data Governance retires with support ending Nov 12, 2027; Purview keeps security and compliance.
  • Microsoft now governs through the OneLake catalog in Fabric; Unity Catalog governs your Databricks lakehouse.
  • A mirrored Azure Databricks catalog in Fabric reads Unity Catalog data without copying it.
  • Atlan adds the cross-platform layer for glossary, ownership, and lineage beyond both.

Are Purview and Unity Catalog competitors?

No. Microsoft announced on November 12, 2026 that Purview Data Governance will retire, with support ending November 12, 2027. The OneLake catalog in Fabric is now Microsoft's data governance solution, and Purview stays focused on security and compliance. Unity Catalog governs your Databricks lakehouse. Most Databricks-on-Azure teams pair Unity Catalog with the OneLake catalog, keep Purview for sensitivity labels and Data Loss Prevention, and add a cross-platform layer when the estate spans more clouds.

Ready to see how Atlan compares?

See Context Layer in Action

Use Unity Catalog inside Databricks, the OneLake catalog for Microsoft governance, and both if you run Databricks on Azure. The comparison changed on November 12, 2026, when Microsoft announced that Purview Data Governance will be retired, with support ending November 12, 2027. Microsoft Purview stays focused on security and compliance, and Microsoft Fabric, with the OneLake catalog, is now Microsoft’s data governance solution. So the useful question is how Unity Catalog, which governs your Databricks lakehouse, fits next to the OneLake catalog for governance and Purview for security.

Map Your Purview and Unity Catalog Coverage


Give it every system in your estate and which catalog governs each today. It returns a coverage table and the specific gap ranked by what breaks first. Read the skill.

Paste into a new chat

Use the skill at https://atlan.com/skills/catalog-coverage-gap-map.md to map catalog coverage across our Purview and Databricks estate. Ask me for whatever it needs.

Run once in a terminal

curl -fsSL --create-dirs \
  -o ~/.agents/skills/catalog-coverage-gap-map/SKILL.md \
  https://atlan.com/skills/catalog-coverage-gap-map.md

For an agent

curl -fsSL https://atlan.com/skills/catalog-coverage-gap-map.md
Factor Microsoft side (OneLake catalog and Purview) Databricks Unity Catalog
Primary scope Fabric, Azure, and third-party sources through shortcuts, mirroring, and connectors Databricks lakehouse plus federated sources
Target users Fabric admins, data owners, compliance and security teams Data engineers, data scientists, ML engineers
Governance model OneLake catalog governs; Purview adds sensitivity labels and DLP Access enforcement at the compute layer
Data discovery Explore tab, embedded in Teams, Excel, Copilot Studio Catalog Explorer in Databricks
Lineage depth Item-level lineage and impact analysis in Fabric Automatic lineage across Databricks assets
Cost basis Governance included in Fabric capacity; Purview labels and DLP can need separate licensing Managed service plus an open-source implementation
Retirement Purview Data Governance support ends Nov 12, 2027 Not affected
Best for Microsoft-ecosystem governance and protection Databricks operational governance, ML and AI, lakehouse security

Why does the comparison between Purview and Unity Catalog change in 2026?

Until November 2026, Purview’s Unified Catalog covered the estate and Unity Catalog covered the lakehouse. Microsoft has now split that role.

What Microsoft announced


Per Microsoft’s November 12, 2026 announcement, Purview Data Governance will be retired. Support ends November 12, 2027. The OneLake catalog is “the single foundation for governing data across your entire estate.” Purview stays focused on classifying, labeling, and protecting sensitive data and preventing data loss.

Microsoft Learn adds detail. Unified Catalog and classic Purview data governance capabilities (formerly Azure Purview) are being retired. New accounts can no longer be created. Existing capabilities stay in maintenance mode through that date.

What it means for Databricks teams


For a team running Databricks on Azure, the comparison moves. The Microsoft side is now the OneLake catalog for governance plus Purview for security and compliance. Unity Catalog sits beside both.

The retirement guide covers the dates. The migration plan covers what teams leaving Purview Data Governance need to carry over.


What are the OneLake catalog, Microsoft Purview, and Databricks Unity Catalog?

Microsoft’s governance now spans two products with different jobs, and Unity Catalog is Databricks’ governance layer for lakehouse data and AI assets. Each covers a different scope, so Databricks-on-Azure estates often run them side by side.

Microsoft: OneLake catalog for governance, Purview for security


The OneLake catalog is the central experience in Microsoft Fabric for discovering data, understanding governance status, and managing security. It has Explore, Govern, and Secure tabs. It is embedded in Microsoft Teams, Excel, and Copilot Studio, per Microsoft Learn.

Microsoft says it governs data across Fabric, Azure, and third-party sources through shortcuts, mirroring, and connectors, without copying data. Governance is included in Fabric capacity rather than billed through separate meters. The Fabric governance guide covers what it includes.

Microsoft Purview began as Azure Purview in 2021. Purview Data Governance is the part being retired. Learn describes it as Unified Catalog and Data Map. What stays is security and compliance:

  • Sensitivity labels and classification that extend across the Microsoft ecosystem.
  • Data Loss Prevention (DLP), which detects sensitive information in supported Fabric and Power BI items and can alert and guide users.
  • Protection for data and AI, which Microsoft frames as preventing data loss across the estate and AI. See Purview controls for AI agents.

Teams that use Unified Catalog features should plan for them to move. Those include data products, Critical Data Elements, and governance domains. The Purview Unified Catalog page explains what they do today.

Databricks Unity Catalog


Databricks Unity Catalog is the governance layer for the Databricks lakehouse. Databricks’ documentation says it governs access through privileges, attribute-based policies, and row and column filters. It tracks lineage automatically from source data to models and dashboards. Catalog Explorer handles discovery. It also governs AI assets, and an open-source implementation is available.

Why the catalog is the lakehouse governance layer


In a lakehouse, the catalog is where access rules and lineage meet the data. That is why Unity Catalog enforces access where queries run, inside Databricks. Privileges and filters are defined and enforced in Databricks.

Unity Catalog’s strengths:

  • Automatic column-level lineage across Databricks assets.
  • Fine-grained access control enforced where queries run.
  • Open-source availability, with compatibility for open table formats.
  • AI asset governance alongside tables and views.

Databricks Unity Catalog is a data cataloging, discovery and governance solution

Databricks Unity Catalog. Image by Databricks.


Are Purview and Unity Catalog competitors or complements?

Complements, and the retirement sharpens the split. Unity Catalog enforces access and captures lineage inside Databricks, the OneLake catalog governs the Microsoft side, and Purview handles labels and loss prevention.

Data governance operates across three layers:

  1. Technical enforcement (Unity Catalog). Governs data and AI assets inside the Databricks lakehouse: access control, automatic lineage, and model governance. Primary users are data engineers, data scientists, and analysts working in Databricks.
  2. Governance and protection on the Microsoft side (OneLake catalog and Purview). Governance runs through the OneLake catalog, and protection through Purview’s labels and DLP. Primary users are Fabric admins, data owners, compliance officers, and security teams.
  3. Business stewardship and cross-platform context (an AI Context Platform such as Atlan). Unifies glossary, ownership, and lineage across tools outside Microsoft and Databricks. Primary users are data product owners, business analysts, governance leads, and domain stewards.

How do the Microsoft side and Unity Catalog compare across eight dimensions?

The OneLake catalog and Purview cover governance and protection for the Microsoft estate, and Unity Catalog covers depth inside Databricks. Read “Purview” in older comparisons as two things now: governance in the OneLake catalog, and protection in Purview.

Data discovery


The OneLake catalog’s Explore tab helps users find Fabric items they can access, with filters by domain, and Microsoft says it reaches Azure and third-party sources through shortcuts, mirroring, and connectors. Unity Catalog’s discovery, through Catalog Explorer, is optimized for lakehouse assets within Databricks.

Aspect OneLake catalog (Microsoft governance) Databricks Unity Catalog
Discovery scope Fabric items plus Azure and third-party sources reached through shortcuts, mirroring, and connectors Databricks assets, plus federated external sources
Where users meet it Fabric, Teams, Excel, Copilot Studio Catalog Explorer in Databricks
Target users Fabric users, data owners, governance teams Data teams and analysts in Databricks
Key strength Reach across the Microsoft estate Depth inside the lakehouse

Data lineage


Fabric shows lineage between items in a workspace and an impact analysis for downstream items. Unity Catalog captures lineage automatically from code run in Databricks. When Purview Data Map scans Azure Databricks Unity Catalog, Microsoft Learn says it fetches table, view, and column lineage from notebook runs, and that page now carries the retirement notice for Purview data governance.

Aspect Fabric (OneLake catalog) Databricks Unity Catalog
Lineage granularity Item-level lineage and impact analysis between Fabric items Automatic lineage across Databricks tables, views, notebooks, and dashboards
Capture method Built into Fabric items Captured automatically as code runs
Cross-platform reach Starts where data enters Fabric, or where mirrored catalogs bring it in Strongest inside Databricks
Key strength Impact analysis within the Fabric estate Granular, automatic lineage within Databricks

Security and access control


Purview supplies sensitivity labels and DLP for Fabric data, and Fabric adds workspace roles and data-level controls for SQL analytics endpoints, warehouses, Direct Lake, and KQL databases. Unity Catalog defines access in Databricks with privileges, attribute-based policies, and row and column filters.

Aspect Microsoft side (Fabric and Purview) Databricks Unity Catalog
Core role Labels, DLP, workspace roles, OneLake security roles Access enforcement at the compute layer
Access control model Sensitivity labels, DLP policies, workspace and OneLake roles Privileges, attribute-based policies, row and column filters
Enforcement scope Fabric items and supported Microsoft surfaces Databricks workloads
Key strength Classification and protection across Microsoft data Enforcement where Databricks queries run

AI and ML governance


On the Microsoft side, Microsoft frames Purview’s role as protecting data and preventing data loss across the estate and AI; Purview controls for AI agents covers the detail. Unity Catalog governs AI assets, including models, alongside tables, and works with AI Gateway to govern AI traffic.

Aspect Microsoft side (Fabric and Purview) Databricks Unity Catalog
Key capabilities Sensitivity labels and DLP for data used by AI; OneLake catalog in Copilot Studio Governance of AI assets and AI traffic with AI Gateway
Core distinction Protects data across the Microsoft ecosystem and AI Governs AI within Databricks
When to use You build on Fabric, Copilot, and Microsoft AI services You run AI and ML workloads on Databricks

Data quality


Purview’s data quality rules live in Unified Catalog, which is part of the Purview Data Governance retirement. Teams relying on them should plan where quality checks move; Microsoft’s transition guidance is the place to start. Databricks provides quality and observability on its side, and Unity Catalog’s lineage helps trace an issue upstream.

Dimension Microsoft side Databricks Unity Catalog
Core capability Unified Catalog data quality rules (part of the retirement) Quality signals tied to lineage in Databricks
Planning note Decide where rules move before November 12, 2027 Stays where the data runs

Integration with existing tools


Microsoft’s approach is reach through Fabric: shortcuts, mirroring, and supported connectors. A mirrored Azure Databricks catalog in Fabric, per Microsoft Learn, lets Fabric workloads read data managed by Unity Catalog with no data movement or replication; only the catalog structure is mirrored, and the data is read through shortcuts. Unity Catalog works well inside Databricks and interoperates through open APIs. Each is built around its own ecosystem, so teams running Databricks, Snowflake, and Fabric together add an independent layer.

Dimension Microsoft side Databricks Unity Catalog
Integration philosophy Reach through Fabric shortcuts, mirroring, and connectors Depth in the lakehouse, open APIs
Native ecosystem Fabric, Azure, Microsoft 365 Databricks
Unity Catalog link Mirrored catalog reads Unity Catalog data in Fabric Source of the mirrored catalog

Cost and pricing


Microsoft says OneLake catalog governance is included in Fabric capacity rather than billed through separate meters, and Purview capabilities such as labels and DLP can need separate Purview licensing, per Microsoft Learn. Purview Data Governance billing ends with the product. On the Databricks side, an open-source implementation of Unity Catalog is available, and Databricks documents the managed service separately.

Deployment and management


The OneLake catalog is part of Fabric, so there is no separate service to provision. Unity Catalog runs inside Databricks. Effort scales with how many sources and teams you connect, and with how much of Purview Data Governance you used.


Can Unity Catalog and Purview work together?

Yes, with a changed shape. The old Purview path is going away, and a Fabric path replaces it.

The Purview scan path


Until the retirement, Purview’s Data Map could scan Azure Databricks Unity Catalog metastores. It brought catalogs, schemas, tables, views, columns, tags, and lineage into Purview. Microsoft Learn now carries the retirement notice on that page. New designs should not depend on it.

The Fabric mirroring path


The Microsoft-side path that does not rely on Purview Data Governance is the mirrored Azure Databricks catalog in Fabric. Fabric workloads read Unity Catalog data in place, and the mirrored items are ordinary Fabric items. Purview’s labels and DLP stay for protection.

When to use Unity Catalog, the OneLake catalog, or both


Pick by scenario:

Scenario Recommended approach Why
Fabric and Azure only, no Databricks OneLake catalog plus Purview security Microsoft’s governance solution, included in Fabric capacity
Databricks only, no Azure requirements Unity Catalog only Native governance inside the lakehouse
Databricks and Fabric (most Azure enterprises) Unity Catalog plus OneLake catalog, with a mirrored catalog Unity Catalog enforces in Databricks; Fabric reads mirrored data
Multi-cloud plus Databricks Both, plus Atlan One glossary, ownership model, and lineage graph across clouds
Currently on Purview Data Governance Plan the move before Nov 12, 2027 See the migration plan and Purview alternatives

What a practical Databricks and Microsoft run-both setup looks like


Teams running Databricks next to Fabric describe a specific seam. Unity Catalog governs the Databricks side. The OneLake catalog governs the Fabric side. Each control plane has its own rules, so someone has to reconcile them. The seam is structural.

  • One system governs a table at a time. Mirroring brings Unity Catalog structure into Fabric, but Unity Catalog’s privileges and filters stay in Databricks.
  • Some governance objects stay in their own system. Unity Catalog’s privileges, row and column filters, and AI Gateway policies are enforced on the Databricks side.
  • Classifications are defined per system. Check how a Fabric sensitivity label and a Unity Catalog tag relate in your own setup before assuming they match.
  • Lineage follows each platform. Unity Catalog’s lineage covers what runs in Databricks, and Fabric’s covers Fabric items. A pipeline that starts in Fabric, lands in Databricks, and reports out through Power BI needs a layer that joins both.

A layer that connects to both on their own terms closes the seam, with no integration work between the two control planes. Atlan’s Databricks connector crawls Unity Catalog schemas, tables, views, volumes, and AI models and builds column-level lineage automatically, and Atlan’s Fabric connector, generally available since version 2.0 in August 2026, catalogs Fabric items down to column level with end-to-end lineage. One business glossary sits above both, so a Databricks table and a Fabric or Power BI asset can carry the same term. Neither system has to change.


Where do teams add a layer beyond the Microsoft side and Unity Catalog?

Each platform governs its own ecosystem. Teams that need lineage, one business vocabulary, and governance workflows across both add a third-party catalog. Microsoft itself says “some organizations have needs that reach beyond what’s in Fabric today.”

Need What spans both Microsoft side (Fabric and Purview) Databricks Unity Catalog
Ecosystem boundaries Governance beyond the native ecosystem Reach through shortcuts, mirroring, and connectors; third-party tools cover sources outside them Governs the Databricks lakehouse; other tools cover sources outside it
Cross-platform lineage End-to-end lineage across both systems and beyond Lineage within Fabric, from where data enters it Lineage within Databricks, from where data enters it
Business context One glossary and ownership model across platforms Domains and endorsement inside Fabric Tags and discovery inside Databricks
Retirement planning A home for Purview Data Governance assets Support ends Nov 12, 2027 Not affected

What happens when governance tools run disconnected?

Disconnected governance tools tend to produce four failure patterns: compliance blind spots, stale metadata, cross-platform silos, and technical compliance without business alignment.

  • Unity Catalog without a Microsoft-side view: security teams need a way to see which labels and policies apply to Databricks data.
  • Microsoft governance without Unity Catalog: the Fabric side sees mirrored structure. Databricks privileges, filters, and in-Databricks lineage stay in Databricks.
  • Either side without cross-platform context: you get metadata silos. Most stacks also include Snowflake, dbt, Tableau, Looker, Airflow, Fivetran, or SaaS apps.
  • Governance tools without a business glossary: access controls work and lineage is captured, but business users still need help finding data and understanding field meanings.

Which criteria should drive the decision?

Pick by where your data lives and who needs to govern it.

  • Where the data runs: Databricks-only estates start with Unity Catalog. Fabric and Azure estates start with the OneLake catalog.
  • Who enforces access: Unity Catalog enforces in Databricks. Fabric roles and Purview labels cover Fabric items.
  • Protection needs: keep Purview for sensitivity labels and DLP.
  • Reach across clouds: add a cross-platform layer when glossary, ownership, and lineage must span more than these two.
  • Purview Data Governance use: plan the move before November 12, 2027.

FAQs: Purview vs. Databricks Unity Catalog

Quick answers to the questions teams ask most. Each follows Microsoft’s November 12, 2026 announcement.

1. Are Microsoft Purview and Databricks Unity Catalog competitors?


No. Unity Catalog governs your Databricks lakehouse. Microsoft’s governance now runs through the OneLake catalog in Fabric, and Purview stays focused on security and compliance. Purview Data Governance is retiring, with support ending November 12, 2027.

2. When should you use the Microsoft side only vs. Unity Catalog only?


Use the OneLake catalog and Purview security alone when your estate is Microsoft-centric and you use no Databricks. Use Unity Catalog alone when all your governed data lives in Databricks and you need no cross-cloud governance. Estates running both usually pair Unity Catalog with the OneLake catalog and add an independent layer for multi-cloud needs.

3. What changed when Databricks open-sourced Unity Catalog?


Databricks documents Unity Catalog as available in an open-source implementation, so organizations can use it outside the managed service. The open-source release also lets other vendors and engines integrate with its governance model.

4. What does the Purview Data Governance retirement change for Unity Catalog teams?


Purview Data Governance, which includes Unified Catalog and the Data Map, is retiring, with support ending November 12, 2027. Scanning Unity Catalog into Purview’s Data Map is part of what retires. Teams should plan to govern Databricks-related assets through the OneLake catalog, Unity Catalog, or a third-party layer, and keep Purview for labels and loss prevention.

5. How do Purview and Unity Catalog work together?


Until the retirement, Purview could scan Unity Catalog metastores into its Data Map. The path that stays supported is a mirrored Azure Databricks catalog in Fabric, which lets Fabric read Unity Catalog data without copying it, while Purview continues to provide sensitivity labels and DLP. Many teams add Atlan to join glossary, ownership, and lineage across both.


How does Atlan extend the Microsoft side and Unity Catalog?

Even with Fabric governance and Unity Catalog in place, many teams still want more for multi-cloud estates, business-user governance, and cross-system lineage. Microsoft’s announcement names Atlan among the third-party providers Fabric integrates with.

Coverage across both platforms


Atlan connects to more than 100 sources, including Databricks and Fabric. Discovery and lineage span platforms in one graph. Atlan’s Context Lakehouse stores context as open Apache Iceberg tables in OneLake that Fabric can query directly. That context reaches people in Microsoft Teams and agents in Copilot Studio. Atlan is available in Microsoft Marketplace.

For teams leaving Purview Data Governance


Atlan’s Migration Studio is in private preview. It reads Purview Data Governance with read-only access and checks the plan against Atlan before writing. The governance lead approves where objects land. A reconciliation report then compares what landed against the plan.

Start with the migration plan. Compare options in Purview alternatives. See coexist, extend or replace Purview for the decision. Best data catalogs for Microsoft Fabric lists the field.


Sources

  1. A next step in the evolution of the Microsoft data governance strategy, Microsoft Security Blog, 2026-11-12
  2. Learn about data governance with Microsoft Purview, Microsoft Learn
  3. OneLake catalog overview, Microsoft Learn
  4. Governance and compliance in Microsoft Fabric, Microsoft Learn
  5. Microsoft Fabric mirrored catalog from Azure Databricks, Microsoft Learn
  6. Connect to and manage Azure Databricks Unity Catalog in Microsoft Purview, Microsoft Learn
  7. What is Unity Catalog, Databricks documentation (docs.databricks.com, data governance, Unity Catalog)
  8. Databricks connector, Atlan documentation

Share this article

signoff-panel-logo

Atlan is the Context Layer for AI, a Leader in the Gartner Magic Quadrant for D&A Governance (2026) and the Forrester Wave for Data Governance (Q3 2025). Atlan unifies your data, business knowledge, and the meaning behind your terms into one Enterprise Data Graph that gives every team and every AI agent the trusted context they need. Trusted by Mastercard, Workday, General Motors, CME Group, HubSpot, FOX, Virgin Media O2, Elastic, and 400+ enterprises representing $10T+ in market cap. In Atlan's AI Labs benchmark, adding that context improved AI's text-to-SQL accuracy by 38%.

Bridge the context gap.
Ship AI that works.