What Is a Composite Model in Power BI?
A composite model is a Power BI semantic model that combines more than one storage mode in a single model. Instead of forcing every table to be either Import or DirectQuery, a composite model lets each table use the mode that fits it best, and it can pull from several sources at once. A large fact table can stay in DirectQuery so it is always current, while smaller lookup tables are imported into memory for speed. This flexibility is why composite models are a common pattern for larger, mixed-source reporting.
How Composite Models Work
In a composite model, Power BI assigns a storage mode per table rather than for the whole model. When a visual needs data, the engine works out which tables are involved and resolves each one according to its mode: imported tables answer from memory, DirectQuery tables send a live query to the source. Relationships can cross modes, so an imported dimension table can filter a DirectQuery fact table. The model stitches the results together so the report behaves as one connected dataset, even though the data lives in different places.
Storage Modes in a Composite Model
A table in a composite model can be set to one of three storage modes. Import keeps a compressed copy in memory for fast reads. DirectQuery leaves the data in the source and queries it live. Dual is a hybrid: the table acts as imported when a query can be served from memory and as DirectQuery when it is joined to a live table, which avoids slow round-trips for shared dimension tables. Setting dimensions to Dual and large facts to DirectQuery is a frequent, effective combination.
When to Use a Composite Model
Composite models earn their place when a single storage mode does not fit the whole dataset. Common cases: a fact table too large or too volatile to import sits in DirectQuery while the rest of the model is imported for speed; data from two systems, such as an ERP warehouse and a separate planning source, needs to live in one model; or aggregations sit on top of a DirectQuery detail table to answer summary questions from memory and fall back to the source only for detail. The cost is added design care, since mixing modes makes performance and refresh behavior harder to reason about.
Composite Models on a Governed Data Foundation
A composite model is only as sound as the data it sits on. Mixing modes across raw ERP tables and ad hoc sources tends to produce models that are hard to maintain and slow to trust. QuickLaunch builds governed data foundations for JD Edwards, Vista, NetSuite, and OneStream with pre-built, well-modeled semantic models, so most enterprise Power BI reporting runs cleanly in Import mode and a composite design is a deliberate choice for the few tables that need live data, not a patch over a messy source.
Frequently Asked Questions
What Is a Composite Model in Power BI?
A composite model is a Power BI semantic model that combines more than one storage mode, so each table can be Import, DirectQuery, or Dual, and the model can draw from several sources at once. It lets large or live tables stay in DirectQuery while smaller tables are imported for speed.
What Is Dual Storage Mode?
Dual is a storage mode where a table behaves as imported when a query can be answered from memory and as DirectQuery when it is joined to a live table. It is often used for shared dimension tables to avoid slow round-trips back to the source.
When Should I Use a Composite Model?
Use one when a single storage mode does not fit the whole dataset: a fact table too large to import, data from two systems that must live in one model, or aggregations layered over a DirectQuery detail table. It adds design complexity, so use it deliberately rather than by default.
