DIMENSIONAL DATA MARTS AND CUBES

Data Marts

ComFrame's BI methodology is based on the rapid creation of business value through incremental delivery of independent data marts. Data marts are analytical data structures, associated with a specific business process, that facilitate storage of historical data, in a business (not technical) context. 

Measures are at the core of every ComFrame data mart. These are generally numeric values associated with a business process or functional activity. For example, a retail sales data mart may have measures that include:

• Units sold
• Revenue
• Number of sales transactions (customer count)

Hierarchies serve to organize groups of dimensions. For example, all activities for the calendar date representing January 1, 2010 would be included in aggregate totals for the first week of 2010, the entire month of January, 2010, the first quarter of 2010 and the entire year of 2010.  Hierarchies make possible the drilldown analysis by users to break down measures into their smallest atomic contributing measure.

• Dimensional data marts provide the added advantage of allowing incremental delivery of business value over time. BI projects no longer require an enterprise scope of modeling and construction to be successful. Instead, the enterprise data warehouse can be constructed over time by the release of individual, independent data marts. 
• In addition to delivering incremental business value through the release of dimensional data marts, ComFrame’s BI methodology provides the following benefits:
  • Predictable schedule, by decomposing the enterprise data warehouse into smaller projects of discrete, business-aligned functionality, software release risk is minimized.
  • Additional functional areas can be incorporated with minimal additional effort.

Cubes

OLAP cubes are logical objects, physically implemented in binary files that provide information consumers with fast analysis of dimensional models.

• Cubes enable you to provide information to various reporting formats (traditional, scorecards, dashboards, and charts) as well as analytical tools for exploratory (slice-and-dice) functionality.
• Cubes must have content updated through a technique known as processing.
• Cube processing must occur after the ETL processes run.