Data Marts

Many Microsoft customers have preferred to implement data marts, which are a type of data warehouse with the following characteristics:

Subject oriented
Fewer dimensions
Limited data feeds
Quicker to implement (average from 3 to 6 months)
Easily accessible
Lower cost
Lower risk
Iterative development (allows prototyping)
Homogeneous group of users who share common business problems, metrics, or roles

Data marts are often specific to an organizational department, unit, or function. Data warehousing is a term that usually is applied to a decision support system that models an entire entity's business and data structure. Some data marts eventually grow or merge into a data warehouse. Another approach is for large enterprises to break a large data warehouse into several data marts, which send reports back to the central data warehouse repository. Or the central data warehouse can feed the data marts their information, in which case the data marts inherit from the data warehouse its business rules, sources, and semantics.