Live Connection & Composite Models
Power BI · Getting Data
Theory
A live connection connects a report to an existing model instead of building a new one: a published Power BI semantic model or an Analysis Services model. The report has no Power Query and no tables of its own; all tables, relationships and measures come from the shared model, though you can add report-level measures. This is how one certified model feeds many reports.
A composite model combines storage modes or sources in one model, for example imported Customers with a DirectQuery Sales table, or a published semantic model extended with your own local Targets table. When you are live-connected to a Power BI semantic model, choosing 'Make changes to this model' converts the connection into a DirectQuery link to that model so you can add tables. Tables can also be set to Dual so they serve both imported and DirectQuery queries.
Why we use it
Live connections stop every analyst from rebuilding the same model, so everyone uses the same definition of Total Sales. Composite models add flexibility when one source or one storage mode is not enough.
When to use it
Use a live connection when a governed shared model already has what you need. Use a composite model when you must add local data to a shared model or mix Import and DirectQuery. Avoid composite designs when one simple Import model would do; they add complexity.
How it works
- Need a new report
- Does a shared semantic model already have the data? Yes: Use a live connection to it. No: Build your own Import model.
- Need extra local data on top? Yes: Make changes to this model (composite). No: Keep the pure live connection.
- Add report-level measures
- Report built on governed data
Syntax
Live: Home → Get data → Power BI semantic models → pick model → Connect
Extend: Modeling → Make changes to this model (creates a composite model)
Storage mode per table: Model view → Properties → Storage modeExample
1. Connect to the certified 'Sales Model' semantic model (live connection).
2. Add a report measure: Sales per Order = DIVIDE([Total Sales], DISTINCTCOUNT(Sales[OrderID]))
3. Need region targets that are not in the shared model? Choose 'Make changes to this model'.
4. Import a local Targets(Region, Target) table and compare it with sales:
Region Target = CALCULATE(SUM(Targets[Target]), TREATAS(VALUES(Sales[Region]), Targets[Region]))
5. The report is now a composite model: shared model via DirectQuery + local imported Targets.Key points
- Live connection reuses a published model; no Power Query in the report
- You can still add report-level measures on a live connection
- Composite = mix of storage modes or sources in one model
- Dual storage mode helps dimension tables serve both modes
- Shared models give one version of the truth across reports