DP-900 · 4 Describe an analytics workload
Power BI data models
Exam objective: Describe features of data models in Power BI
A semantic model defines measures, the numbers you analyze, and dimensions, the entities you group them by, linked in a star or snowflake schema.
Before a report can show anything useful, its data has to be organized into a model, called a semantic model in Fabric and Power BI, that defines what you measure and what you group by.
A measure is the numeric value you care about, such as revenue or quantity sold, and it lives in a fact table where each row is one recorded event, like a single order. A dimension is an entity you aggregate measures by, such as product, customer or time, and it lives in a dimension table with a unique key per row plus descriptive attributes. A fact table linked to one or more dimension tables is a star schema. If a dimension table is further linked to additional detail tables, for example splitting Product off into a separate Category table, the result is a snowflake schema.
Power BI loads this model into an in memory, columnar store called the VertiPaq engine, and computes aggregations at query time rather than storing them in advance. Hierarchies, such as category under product, or year, month and day within a time dimension, let users drill up and down through those levels in a report. Direct Lake mode connects a model straight to OneLake files for similar in memory speed without importing the data first.
On the exam, a numeric value to analyze is a measure, and an entity to group by is a dimension.
Key points
- A measure is a numeric value to analyze; it lives in a fact table with one row per recorded event.
- A dimension is an entity to aggregate by; it lives in a dimension table with a unique key and descriptive attributes.
- A star schema links a fact table to one or more dimension tables; a snowflake schema extends a dimension table with further detail tables.
- The VertiPaq engine stores a loaded semantic model in memory and computes aggregations at query time.
- A hierarchy lets a report drill up and down through levels within a dimension, such as year, month and day.
Exam trap
Fact and dimension tables get swapped often. The fact table holds the numbers you are measuring, one row per event, while the dimension table holds the descriptive entities you slice those numbers by.
Check yourself
A Product dimension table is linked directly to a sales fact table. The Product table is also linked to a separate Category table that holds category details. What is this design called?
Go deeper on Microsoft Learn
Checked against Microsoft Learn on October 1, 2026.