Power BI storage modes: decide where each table keeps its data
Power BI storage modes control where each table in a semantic model keeps its data and how queries reach it. Some tables hold a cached copy, others ask the source every time, and some do both. The choice shapes speed, freshness and model size. This guide explains each mode, where you find the setting and the changes that Microsoft says you cannot reverse. In short, Import is fastest, DirectQuery is freshest, Dual serves both, and Microsoft does not let you switch an Import table to another mode.
- Import, DirectQuery, Dual
- Set per table
- Some changes are one-way
Last updated
What are Power BI storage modes?
Microsoft describes the storage mode as a table property. It determines how and where the table stores the data that reports show. So one model can treat its tables differently, and you decide table by table.
An Import table stores a snapshot in Power BI's own storage. You refresh the model to bring in newer data. A DirectQuery table stores nothing. Instead, Power BI queries the source when a report visual loads and translates the DAX query into the language of that source.
When a model contains tables with more than one storage mode, Microsoft calls it a composite model. A live connection is a different thing. Such a report has no local model and only connects to a published one.
Power BI storage modes compared side by side
Microsoft's article lists these modes, and the notes give a short version of each description.
| Mode | Where the data sits | Documented note |
|---|---|---|
| Import | A snapshot in Power BI storage | Refresh the model to get newer data |
| DirectQuery | Only in the source | Queried when a report visual loads |
| Dual | Cached, and also in the source | Acts as Import or DirectQuery per query |
| Direct Lake on OneLake | In Fabric data sources | Added through the OneLake catalog |
| Direct Lake on SQL | Behind a SQL analytics endpoint | Can fall back to DirectQuery |
| Hybrid | Partly cached, partly in the source | Used with incremental refresh |
Check a table's mode in four steps
These steps follow Microsoft's description for Power BI Desktop.
Open Model view
First switch to Model view in Power BI Desktop, where the tables and relationships appear as a diagram.
Select the table
Next select the table you want to inspect. The Properties pane then shows its settings.
Expand Advanced
Then expand the Advanced section of the Properties pane and find the Storage mode dropdown.
Read before you change
Finally read the current value. However, check the one-way rules below before you pick another mode.
Why does the Dual mode exist among Power BI storage modes?
Dual answers a relationship problem. Suppose a fact table uses DirectQuery and an aggregated table uses Import. A shared dimension table then has to work with both. As a Dual table, it can act as either one, depending on the query.
Microsoft explains that this reduces the number of limited relationships in the model and improves performance. Power BI Desktop even helps with the setup. When you change a table, it works out the smallest set of dimension tables that must become Dual and offers to set them.
There is a trade-off, however. Queries hit the cache when possible, and cached values can differ from the source until the next refresh. Microsoft treats Dual as a performance optimization, and it makes the model author responsible for consistent results.
Which storage mode changes are one-way?
Usually you set the mode only when you add the table. Microsoft documents one later change: a DirectQuery table can become an Import or a Dual table. That step is not reversible in Power BI Desktop, so you cannot switch the table back.
Two exceptions exist. Web modeling and live editing of semantic models have version control, and Microsoft notes that you can go back there. For everything else, keep a copy of the file before you convert.
Also check the source. According to the article, SAP HANA and SAP Business Warehouse sources that are treated as multidimensional cannot take part in a composite model. Dual tables carry the same functional constraints as DirectQuery tables as well.
Three questions to ask before you pick a mode
No single mode fits every table, so weigh these points for each one.
How fresh?
Import shows data as of the last refresh. DirectQuery asks the source each time, so visuals reflect its current state.
How large?
Import stores a copy, so the model grows with the data. DirectQuery keeps the data in the source instead.
How shared?
A dimension that filters both cached and DirectQuery tables is a candidate for Dual, as Microsoft's example shows.
Power BI storage modes: common questions
Where do I see Power BI storage modes for a table?
Open Model view in Power BI Desktop and select the table. Then look in the Properties pane, expand the Advanced section and read the Storage mode dropdown. Microsoft also notes that Table view shows data only for Import and Dual tables, so a DirectQuery table does not appear with its rows there.
Can I change Power BI storage modes after the table is added?
Only in one direction inside Power BI Desktop. A DirectQuery table can change to Import or Dual, and that change cannot go back. Microsoft lists web modeling and live editing of semantic models as exceptions, because they have version control. So save a copy of the file before you convert a table.
What is a composite model?
It is a semantic model whose tables use more than one storage mode. For example, a large sales table can stay in DirectQuery while small lookup tables use Import or Dual. Microsoft documents some limits. SAP HANA and SAP Business Warehouse multidimensional sources cannot be part of such a model.
Which of the storage modes in Power BI is fastest?
Microsoft's article gives no ranking, so we will not invent one. It does state that Dual tables can answer from the cache when possible and that this improves performance in composite models. In practice the answer depends on the data volume, the source and how fresh the figures must be.
Can Power BI Agency choose the right setup for our model?
Yes. One of our consultants reviews each table, its size and how current it must be, then recommends a mode for it and explains the trade-offs. We work online in your own Microsoft tenant and begin with a free 30-minute assessment. Power BI Agency is a division of Sataware Technologies, independent of Microsoft.
Let's build your first dashboard together.
Tell us what you report on today. A consultant replies within 2 hours with next steps.
Trusted partners
Microsoft Solutions Partner · AWS Partner
Cloud & AI Platforms
Azure SQL, Microsoft Fabric and the data pipelines behind your reports.
AI Business Solutions
Power BI reporting on your Dynamics 365, Business Central and Power Platform data.
Security
Row-level security and Entra ID sign-in on every report and workspace.
Amazon Web Services
Power BI on your AWS data, with connectors for Redshift, Athena and S3.