Comparison of two data warehouse design architectures

Bill Inmon and Ralph Kimball, who were exposed to two names at school, were unfamiliar to most of the two Americans, but they were a resounding figure in the database field. Bill Inmon, known as the "Father of the Data Warehouse", he can now see a lot of scholarly papers and articles on the Web, and Wikipedia's introduction to him should be very comprehensive: in the 80 's, Inmon's "Data Warehouse" book defines the concept of data warehousing, A more precise definition is given: The Data Warehouse is a topic-oriented, integrated, time-dependent, and unmodified collection of data in enterprise management and decision-making. Unlike other database applications, a data warehouse is more like a process for consolidating, processing, and analyzing business data that is distributed throughout the enterprise. Rather than a product that can be purchased.

Ralph Kimball, like Bill Inmon, even proposed the concept of a data warehouse earlier, except that he had only proposed a bottom-up architecture, and Inmon summed up a series of theories. The Inmon and Kimball two DW architectures support the development of data warehousing and business intelligence for nearly 20 years, where Inmon advocates Top-down architecture, with different OLTP data focused on themes, integrated, volatile, and time-changing structures for future analysis; And the data can be drilled to the finest layer, or rolled up to the summary layer; The data mart should be a subset of the data Warehouse; Each data mart is specially designed for an independent department. And Kimball is just the opposite of Inmon, the Kimball Architecture is a bottom-up architecture that considers data warehouses to be a collection of data marts. Organizations can incrementally build data warehouses through a series of data marts of the same number of dimensions, using consistent dimensions to see information in different data marts together, indicating that they have publicly defined elements.

As you can see, there are two completely different architectures, and in the early days, the Kimball approach got more applications, and Microsoft Business Intelligence used the Kimball architecture. As technology continues to evolve, the two architectures of Kimball and Inmon are becoming more similar, and they are struggling in the forefront of data warehousing, and business intelligence advocates deserve it.

A simple comparison of the two architectures:

Inmon: Don't do any work until you design all the components

Advantages: Easy to maintain, highly integrated

Disadvantages: Rigid structure, long deployment cycle

Kimball: As long as users have needs, they can build the components they want and then integrate them.

Advantage: Build quickly, see ROI quickly, agile and flexible

Disadvantage: As enterprise resources are not very good maintenance, complex structure, data mart integration difficulties

(Responsible editor: Lu Guang)