Two major schools of Data Warehouse theory (_bi.old)

The Kimball of Master Ralph:

The Inmon of Bill master

1. Brief introduction

The architecture design of BI system is the core of Data Warehouse architecture design. In the field of Data Warehouse architecture, there are two schools of theoretical knowledge, which were presented by two Masters Ralph Kimball&bill Inmon in the early 90. These two masters are the major gurus and theorists in the field of business intelligence/data warehousing, but their two-bit ideas and ideas are quite different. There is also a constant debate among their followers about which architecture and how to build it.

At home, what we normally call the four-point definition of a data warehouse (theme-oriented, integrated, relatively stable, reflective of historical change, used to support decision making) is Inmon, who is also known as the father of the Data warehouse. And the Practice Master Kimball his Toolbox series of works, also is the data storehouse construction classics.

2. Data Warehouse Architecture Features

Kimball supports data Warehouse bus structure, advocates dimension modeling, and constructs dimension Data Warehouse in the form of star model or snowflake model, in which data mart is closely combined with dimension Data Warehouse, and data mart is a logical subject domain in Data Warehouse. The various front-end tools will have direct access to the use of the dimension Data Warehouse.

Inmon advocates the construction of centralized enterprise Data Warehouse (EDW) in three normal forms, as the core of the whole system, and then constructs several theme-oriented data marts that follow the dimension model design. In this system, the front-end tool usually accesses the data mart without directly accessing the Edw.

3. Characteristics of construction methods

The

Kimball is recommended from the bottom up. First, a number of data marts are built to meet the needs of departmental analysis, then integrated with the bus architecture to form a "federated Data Warehouse". This method is often referred to as the bottom-up approach. The
Inmon emphasizes the self entry downward. First, the data from each source business system is integrated into the enterprise-level data Warehouse, and then based on the data mart which needs to be used by the department.
4. How to choose each applicable scene.
Starting        kimball                       &N Bsp                inmon
Planning level     tactical           &N Bsp                                 Strategy
Data integration     Meet part of the specified requirements can be                 Enterprise-wide data integration
Technology-related     Small team, skill requirement Like         need large team, high skill requirement
Time constraint     have urgent requirements               &NB Sp               allow long period construction
construction cost     lower cost implementation           &NB Sp                   high cost

The Kimball method is not very high to the team skill level, it is easier to realize, from the small subject area data mart construction, but in the gradual construction process, the consistency of the joint Dimension Data Warehouse is more difficult to control, it is suitable for the technology level planning, or has the urgent goal needs to realize.
Inmon, the standard is better, data mart and data consistency can be dealt with, suitable for larger enterprise-level, strategic-level planning, but the requirements of the team is high, and the implementation cycle is long and expensive.

Specific can be based on the size of the enterprise, project planning, budget, team and other aspects of the comprehensive consideration.

--from TS Video