20.2. SSAS Overview
A data warehouse combines data from multiple sources into a single homogenous repository that is organized for efficient analytical query processing rather than transaction processing. Data warehouses use dimensional modeling to represent business information. Data is stored in two types of tablesfact tables and dimension tables .
Fact tables contain numeric performance information from transactional data. The columns of a fact table are one of two typesmeasures or attributes. A measure is quantitative business data and is usually numeric. An attribute is used to associate the measures with a row in the dimension table.
Multidimensional data is represented by structures called cubes, each representing a set of data called a measure, hierarchically organized by one or more dimensions. Dimensions organize data within a cube by using hierarchies and attributes instead of tables. Cubes are typically built from data in relational data sources. A member is an item in a dimension that represents one or more data instances, similar to records in a relational database.
A hierarchy organizes the members of a dimension into one or more levels, and lets you navigate and aggregate data within the cube. Each attribute in a hierarchy definition corresponds to a level in the hierarchy from the most summarized down to the most detailed. For example, you can have a Geography dimension that organizes the hierarchy into levels based on Country, State, and City attributes. The member in the lowest level is called a leaf member, and other members are called nonleaf members.
Multiple data values called measures exist at each intersection of the dimensions. A measure is a special dimension that represents the data organized according to the other dimensions in the cube. Every cube must have a measure dimension. Calculated measures can be created that derive from existing measures.