Star schema

A star schema is a way of structuring data for analysis in which one or more central fact tables, holding measurable events like orders or page views, connect to surrounding dimension tables that describe the who, what, when and where of each fact, such as customer, product or date. Drawn out, the fact table sits in the middle with dimensions radiating outward, giving the pattern its name.

It is the most common form of dimensional data modeling, and differs from the normalized schemas used in transactional databases, where data is split across many linked tables to avoid duplication. A star schema deliberately denormalizes dimensions to minimize joins, which makes it fast for the aggregation and slicing typical of OLAP queries and BI tools. A related variant, the snowflake schema, normalizes dimensions further at some cost to query simplicity.

Star schemas matter because they are easy for both query engines and analysts to reason about: a fact table plus a handful of dimensions maps directly onto how business questions are usually asked. They are widely used inside a data warehouse and are a common output of dbt modeling. Pitfalls include letting dimension tables grow inconsistent across different fact tables, and over-normalizing dimensions in the name of tidiness at the cost of query performance.