You must connect Power BI to a 2 TB transactional database. Users need to explore data that changes continuously, and the source is far too large to load into memory. Which storage mode should you choose?
- AImport
- BDirectQuery
- CDirect Lake
- DDual
DirectQuery stores only metadata and translates DAX into native source queries executed at query time, so it handles data too large to import and data that must stay near real time. Import loads compressed data into the VertiPaq in-memory engine (size limited, refresh fresh). Direct Lake needs Delta tables in OneLake on a Fabric capacity. Dual applies per table for dimensions shared between Import and DirectQuery.
Why each option is right or wrong.
- B (correct). DirectQuery keeps only metadata and sends native queries to the source at query time, suiting very large or near-real-time data.
- A. Import copies and compresses data into memory, so a 2 TB source cannot fit and the data is only as fresh as the last refresh.
- C. Direct Lake requires the data to already exist as Delta tables in OneLake on a Fabric capacity, which is not the case here.
- D. Dual is a per-table setting for a dimension shared between Import and DirectQuery, not a whole-model choice for a huge fact source.
Reference: Storage modes Import vs DirectQuery vs Direct Lake (Prepare the data).