Attributes

Each dimension contains attributes. Each attribute is bound to one or more columns in a dimension table. Attributes can be, and are generally, arranged into user-defined hierarchies that provide the drill-down paths by which users can browse the data in the measure groups to which the attribute is related. You might also have attributes like colors, flavors, or sizes. They are not part of a hierarchy in that you can’t drill up or down into them.

Attributes represent a single type of information in a dimension. For example, quarter is an attribute in the time dimension.

You can use dimension attributes as part of your analysis to tease out more information—for example, if you were researching airline travel habits you could see on which days of the week most people travel for business travel, vacation travel, etc. Or, you can use them to find out which colors were most popular for a particular season or year or garment.

Level Attributes

Level attributes provide additional information about the dimension members at a particular level of a dimension hierarchy. The dimension members themselves may be meaningless, such as a value of “365” for a time period. These values for dimension members are used internally for selecting and sorting quickly but are meaningless to users.

For example, you might have columns for employee number (ENUM), last name (LAST_NAME), first name (FIRST_NAME), and telephone extension (TELNO). ENUM is the best choice for a level column, since it is a key column, and its values uniquely identify the employees. ENUM also has a NUMBER data type, which makes it more efficient than a text column for the creation of indexes. LAST_NAME, FIRST_NAME, and TELNO are attributes.

Dimension Attributes

Dimension attributes specify groupings of level attributes for a specific dimension. Whereas level attributes map to specific data values, dimension attributes are purely logical metadata objects.

An example of a dimension attribute is end date, which is required for time dimensions. If a time dimension has month, quarter, and year levels, end date identifies the last date of each month, each quarter, and each year.