Teradata supports "logical data marts". What does that mean? How is it architected?

A data mart is a pre-selected segment of data needed for a specific application. A physical data mart is an actual recreation of that data segment, typically with some required processing to address the application requirement. A "logical data mart" is a representation of the data needed for a specific application. In Teradata, this is accomplished through views.

Views select data from different tables and potentially include case statements and data translations for specific applications. The use of views eliminates the complication of physical data marts. Additional processing to create the marts is not needed. As data is updated, it is reflected through these views, thus eliminating the need for additional space to store the data mart and making it a "logical" data mart. A logical data mart preserves the data warehousing rule of a single version of the truth.