Page 18 - DCAP603_DATAWARE_HOUSING_AND_DATAMINING

Data Warehousing and Data Mining

notes Data Warehouse Models
From the architecture point of view, there are three data warehouse models: the virtual warehouse,
the data mart, and the enterprise warehouse.
Virtual Warehouse: A virtual warehouse is created based on a set of views defined for an
operational RDBMS. This warehouse type is relatively easy to build but requires excess
computational capacity of the underlying operational database system. The users directly access
operational data via middleware tools. This architecture is feasible only if queries are posed
infrequently, and usually is used as a temporary solution until a permanent data warehouse is
developed.
Data Mart: The data mart contains a subset of the organisation-wide data that is of value to a
small group of users, e.g., marketing or customer service. This is usually a precursor (and/or a
successor) of the actual data warehouse, which differs with respect to the scope that is confined
to a specific group of users.
Depending on the source of data, data marts can be categorized into the following two classes:
1. Independent data marts are sourced from data captured from one or more operational
systems or external information providers, or from data generated locally within a
particular department or geographic area.
2. Dependent data marts are sourced directly from enterprise data warehouses.

Enterprise warehouse: This warehouse type holds all information about subjects spanning the
entire organisation. For a medium- to a large-size company, usually several years are needed to
design and build the enterprise warehouse.
The differences between the virtual and the enterprise DWs are shown in Figure 1.4. Data marts
can also be created as successors of an enterprise data warehouse. In this case, the DW consists of
an enterprise warehouse and (several) data marts.
figure 1.4: a virtual Data Warehouse and an enterprise Data Warehouse

12 LoveLy professionaL university