Microsoft Fabric integration
Common queries around Microsoft Fabric integration. The Microsoft Fabric connector is generally available, and Atlan is available in the Microsoft Marketplace and is Microsoft Azure co-sell eligible.
Atlan crawls Microsoft Fabric workspaces, reports, and dashboards into the catalog; this FAQ answers common integration questions.
Find answers to frequently asked questions about the Microsoft Fabric connector integration.
Does enabling Scanner API Access affect which permissions are required?
No. Both modes need the same permissions: the admin API settings configured in the Fabric Admin Portal, and a workspace role (Viewer or Contributor) on each workspace you want to crawl. This applies to both authentication methods (Service Principal and APIM Managed Identity).
Workspace permissions are needed even when Scanner API Access is enabled, because the Scanner APIs don't return everything Atlan catalogs. Atlan reads lakehouse and warehouse tables and columns from each lakehouse and warehouse directly, and Fabric returns definitions and shortcut targets only to identities with the Contributor role. For what Atlan extracts with each role, see What Atlan extracts by workspace role.
When Extract Spark runtime lineage is enabled, see Set up Spark runtime lineage for its permissions.
Why do semantic models show no tables after crawling?
If every semantic model in one workspace has no tables, while models in other workspaces do, and Scanner API Access is disabled, check the crawl identity's role on that workspace. Atlan reads semantic model definitions with a Fabric API that requires the Contributor role. With Viewer, the crawl still succeeds, but Atlan doesn't extract semantic model tables and columns, report or pipeline definitions, or lineage through those semantic models.
Assign the Contributor role on the workspace and run the crawl again. See Assign workspace permissions.
Does Atlan catalog OneLake shortcuts?
Yes. Tables that appear in a lakehouse through OneLake shortcuts are cataloged as lakehouse tables. When a shortcut points to a table in a Fabric lakehouse or warehouse, Atlan also draws lineage from the source table to the shortcut table, including across workspaces. For this lineage, both workspaces need to be crawled, and the crawl identity needs the Contributor role on the workspace that contains the shortcut.
Atlan doesn't yet draw lineage for shortcuts to storage outside Fabric, shortcuts that bring in a whole schema, or shortcuts to items Atlan doesn't catalog. See Shortcut lineage.
How does Scanner API toggle affect catalog coverage and lineage?
The Microsoft Fabric connector can run in two modes based on the Enable Scanner API Access toggle in the crawler workflow.
- Scanner API enabled (toggle on): Uses the Power BI Admin Scanner APIs for most metadata. This suits centralized admin setups but limits catalog coverage.
- Scanner API disabled (toggle off): Uses both scanner and non-scanner APIs for full catalog coverage.
Both modes require a workspace role (Viewer or Contributor) on each workspace. See What Atlan extracts by workspace role.
| Asset or lineage capability | Scanner API enabled | Scanner API disabled |
|---|---|---|
| Catalog Workspaces | ✅ Available | ✅ Available |
| Catalog Lakehouses, Warehouses (tables, views, columns) | ✅ Available | ✅ Available |
| Catalog Semantic Models (tables, columns, and measures) | ✅ Available | ✅ Available |
| Catalog Reports | ✅ Available | ✅ Available |
| Catalog Dashboards | ✅ Available | ✅ Available |
| Catalog Dataflows (Dataflow Gen1, Dataflow Gen2, and Dataflow Gen2 (CI/CD)) | ✅ Available | ✅ Available |
| Catalog Mirrored databases (tables, views, columns) | ✅ Available | ✅ Available |
| Dataflow last updated date and creator (Dataflow Gen1 and Dataflow Gen2) | ❌ Not available | ✅ Available |
| Catalog Report Pages | ❌ Not available | ✅ Available |
| Catalog Report Visuals | ❌ Not available | ✅ Available |
| Catalog Pipeline Copy Activities | ❌ Not available | ✅ Available |
| Catalog Notebooks and Spark jobs, with their lineage (requires Extract Spark runtime lineage) | ✅ Available | ✅ Available |
| End-to-end lineage (external sources → Semantic Models) | ✅ Available | ✅ Available |
| Lineage: Semantic Model → Report | ✅ Available | ✅ Available |
| Lineage: Semantic Model Table/Column → Report Page or Visual | ❌ Not available | ✅ Available |
| Lineage: source table → OneLake shortcut table | ✅ Available | ✅ Available |
| Lineage: source table → mirrored database table (supported mirroring sources) | ✅ Available | ✅ Available |
| Workspace-level permissions required | ✅ Required (Viewer or Contributor per workspace) | ✅ Required (Viewer or Contributor per workspace) |
For setup instructions specific to each mode, see Set up Microsoft Fabric — Scanner API permissions.
Why fabric semantic model tables don't appear in global Assets search
FabricSemanticModelTable assets are intentionally excluded from Atlan's global Assets search. This is by design—it isn't a bug or an access control issue. Semantic model tables are internal sub-components of a Semantic Model and are designed to be discovered through their parent.
To find a specific semantic model table:
- Search for and open your parent Semantic Model asset.
- Open the Related Assets tab.
- Filter by Semantic Model Table.
Does Atlan catalog Fabric notebooks?
Yes, when Extract Spark runtime lineage is enabled in the crawler. Atlan catalogs each notebook in the crawled workspaces and, from Fabric Spark runtime lineage events, the Spark jobs each notebook runs, with table- and column-level lineage. This requires capture to be set up in Fabric first. See Set up Spark runtime lineage.
Why don't recent notebook runs show lineage in Atlan?
Atlan reads runtime lineage events only for UTC hours that have ended, so a run from the current hour appears after the next crawl that starts once the hour ends. If lineage is still missing, confirm that the tables the notebook reads and writes are cataloged in Atlan. See Troubleshooting Spark runtime lineage.
Can I catalog Microsoft Fabric mirrored databases?
Yes. Atlan catalogs mirrored databases, including mirrored Azure Databricks catalogs, with their schemas, tables, views, and columns. Tables, views, and columns need a workspace role on the mirrored database's workspace. For supported mirroring sources, such as Snowflake, Azure SQL Database, and SQL Server, Atlan also draws table- and column-level lineage from the source tables to the mirrored tables. See Mirrored database lineage.
Known limitations of Microsoft Fabric connector
Atlan currently doesn't support the following Microsoft Fabric features and experiences:
Data Engineering & Analytics:
- Mirrored database upstream lineage - not drawn for SAP Datasphere, open mirroring, or Azure Cosmos DB mirrors. For Google BigQuery mirrors, Atlan matches the source on dataset and table names only, because Fabric doesn't report the project
- Fabric SQL databases and Cosmos DB in Fabric - not cataloged
- Data Science (ML models) - not ingested or linked to upstream/downstream
- Real-Time Analytics (KQL) - not supported as source/lineage entities
- Data Activator - not represented in catalog or lineage
Shortcuts:
- Shortcut labels - shortcut tables are cataloged as lakehouse tables and aren't labeled as shortcuts
- Shortcut lineage - not drawn for shortcuts to storage outside Fabric (such as Azure Data Lake Storage, Amazon S3, or Google Cloud Storage), shortcuts that bring in a whole schema, or shortcuts to Fabric SQL databases, KQL databases, or mirrored databases
- Shortcuts in the Files folder - not cataloged
Dataflows:
- Dataflow Gen2 (CI/CD) without Contributor access - Atlan reads Dataflow Gen2 (CI/CD) definitions through a workspace-level Fabric API that requires the Contributor role. No admin or Scanner API returns them. Without that role, the dataflow is cataloged but has no entity columns or lineage, even when Scanner API Access is enabled
- Dataflow Gen2 (CI/CD) destinations outside Fabric - Atlan catalogs entity columns and downstream lineage only for queries that write to a crawled Fabric lakehouse or warehouse table
- References between queries - lineage doesn't follow one dataflow query referencing another, or queries combined with
Table.Combine
Metadata & Governance:
- Visual Metadata - visuals inside Power BI reports (charts, fields, filters) not captured
- File-Level Metadata - file-level (CSV/Parquet) assets in Lakehouse.
- Governance Metadata - RLS/OLS roles from Fabric not yet ingested
Scanning & Usage:
- Incremental Scans - only full/scoped scans supported; no change-aware deltas
- View in source - not supported
- Usage statistics and Popularity - not supported for Fabric assets