Raw data storage mechanisms differ significantly from the logical structures required for efficient analysis. While the previous chapter addressed the storage layer, this section focuses on the logical organization of data. Dimensional modeling provides a specialized framework for structuring data to support high-performance querying and intuitive reporting.

The primary objective here is to decouple measurement data from descriptive context. You will learn to distinguish between Fact tables, which store quantitative metrics, and Dimension tables, which provide the who, what, where, and when. A central component of this process is defining the "grain" of the data. A clearly defined grain ensures that metrics such as revenue or quantity behave predictably during aggregation. For instance, in a strictly additive fact table, a total across any dimension is calculated simply as $Total = \sum x_i$.

We will examine the specific attributes of Star Schemas, where dimensions connect directly to a central fact table, and contrast this with Snowflake Schemas, which normalize dimension tables to reduce redundancy. Through these lessons, you will acquire the ability to translate business requirements into a schema design that balances query speed with maintenance requirements.