1.2 — Workspace and Item Access, and Data Sharing
DP-600 Domain 1 · Skill 1.2
What You Need to Know
An analytics engineer decides who can build, who can query and who can only view reports. Fabric gives you three layers: workspace roles (everything in a workspace), item permissions (one item) and data-level security (lesson 1.3).
Workspace roles
| Role | Can do |
|---|---|
| Admin | Everything, including deleting the workspace, managing all access, and connecting Git |
| Member | Everything a Contributor can, plus add users (up to Member), share items and allow resharing, and publish apps |
| Contributor | Create, edit and delete items, write data, run notebooks and pipelines, refresh semantic models |
| Viewer | View reports and content, query SQL endpoints with T-SQL. Can't read lakehouse files with Spark or edit anything |
- A role covers every item in the workspace. Assign roles to groups wherever you can.
- Report consumers who should see reports but not the workspace's other items are better served by an app or by sharing the report.
Item permissions
| Item | Permissions you can share |
|---|---|
| Semantic model | Read; Build (create new reports on it, query it with XMLA, analyze in Excel); Write; Reshare |
| Report | Read (view); Reshare. To view, users also need Read on the underlying model, which sharing usually grants |
| Lakehouse | Read (always); Read all with SQL analytics endpoint (ReadData); Read all with Apache Spark (ReadAll); Execute Apache Spark |
| Warehouse | Read (connect only; then T-SQL GRANTs decide what's visible); Read all data using SQL (ReadData); Read all data using Apache Spark (ReadAll); Monitor; Audit |
Build permission is what self-service report authors need on a shared semantic model. Without it they can view existing reports but can't create new ones from the model.
Data sharing patterns
| Need | Pattern |
|---|---|
| Another team needs your lakehouse tables in their workspace without a copy | They create a OneLake shortcut to your lakehouse (and need permission on the source item) |
| Analysts query your warehouse with T-SQL, but only the gold schema | Share the warehouse with Read only, then GRANT SELECT ON SCHEMA::gold |
| Data scientists read lakehouse files in their notebooks | Share the lakehouse with Read all with Apache Spark |
| Business users build their own reports on your certified model | Give them Build on the semantic model (or include it in an app with Build permission) |
| Cross-database queries inside one workspace | Warehouses and SQL endpoints in the same workspace can be joined with three-part names |
| Another organization (tenant) needs your data | External data sharing (OneLake data sharing across tenants, if a Fabric admin has enabled it) |
Key Concept: Roles are about the workspace. Item permissions are about one item: Build for model reuse, ReadData for SQL, ReadAll for Spark. Shortcuts and external data sharing let others use the data without copying it.
Exam Traps
- Making self-service authors workspace Contributors when Build on the model is enough
- Expecting a Viewer to read lakehouse files in a notebook
- Sharing a warehouse with ReadData when only one schema should be visible
- Copying a lakehouse into another workspace instead of creating a shortcut
Check your understanding
Practice Scenario
A certified semantic model lives in the Finance BI workspace. Thirty analysts in other departments must be able to create their own reports on this model. They must not edit the model or see the workspace's other items. What should you grant them?
Log in to save your progress.
Sources
- Roles in workspaces in Microsoft Fabric, Microsoft Learn
- How lakehouse sharing works, Microsoft Learn
- Share your warehouse and manage permissions, Microsoft Learn
- Build permission for shared semantic models, Microsoft Learn
- External data sharing in Microsoft Fabric, Microsoft Learn