Consider the feasibility of two kinds of data Warehouse architecture coexistence

Search for "Inmon and Kimball" on Google, and you'll easily find the concepts of these two names, which are two of the best-known ways of data Warehouse architecture. In this ocean of information, however, you will find that almost all of the content can come to a conclusion, that is, to choose between Bill Inmon and Ralph Kimball.

But the "Father of Data Warehouse" Bill Inmon tells us that in a certain environment, the two architectures can coexist and collaborate well. In this regard, Bill Harrison, bi system architect from the power company, said: "Both architectures have their own applications, and Inmon's standardized data model is great for a centralized data warehouse, but Kimball is better when you design a data mart." So there's no reason why we can't use both. ”

Review of the history of Inmon and Kimball

In a blog techtarget database editor, we compared the two data warehouse architectures of Inmon and Kimball, so here's a review of the history of their two. Bill Inmon and Ralph Kimball published their methodologies early in the 90, but one of them is the same: they want to help companies achieve efficient information management and make better business decisions. It's just a different way of accomplishing it.

Inmon is known as the "Father of Data Warehouse", his method is to build enterprise data Warehouse--Centralized relational database management system, it can provide users with access to high-quality, highly integrated, standardized data capabilities. Interestingly, Inmon was the first to admit that his approach was more expensive and less likely to see ROI in the short term, but Inmon also stressed that his approach had a long lasting return on investment.

Kimball, known as "the Father of business intelligence", pioneered the concept of a data mart, a small information base designed to meet the needs of specific sectors within the enterprise, such as finance, human resources, sales, etc. While Kimball's spatial model provides rapid ROI, experts say it is tricky to maintain data quality in the ocean of data marts.

Both Data Warehouse architectures have undergone continuous evolution over the past 20 years, and the Inmon architecture currently contains a text data warehouse, while Kimball is now more concerned with data consistency. In recent days, TechTarget's website has interviewed Bill Inmon, who believes that the two architectures can be symbiotic and even have unexpected "chemical reactions".

"The Kimball architecture should be good for data marts, and departmental data management can be done by building data marts," Inmon said, "but businesses also need to control data marts through a centrally consolidated data warehouse, which can be used in classic Inmon-mode data warehouses." ”

The proof of coexistence of Inmon and Kimball

In TechTarget's report this week, we shared a case study of the implementation of a Data integration modeling tool. Omaha Power, from the United States, implemented the Inmon-mode data Warehouse and the Kimball-mode data mart, and their bi-system architect Bill Harrison said it was nonsense to report that Inmon and Kimball could not coexist in the network.

"This is not a year or two, the online coverage is more hype, eye-catching," Harrison said: "I think Inmon and Kimball can not coexist the argument is absurd, the two architectures should be complementary, we can use together." ”

Harrison says it makes sense to choose Kimball when building a data mart because it is relatively simple to understand and provides excellent performance and quick returns. This architecture is designed to deliver data to business users as quickly as they can, and most source systems are not designed to do so, instead they are designed to get data from business people faster.

For example, the Omaha Power Company runs Oracle PeopleSoft software to help businesses track customers in specific areas. The information tables in the PeopleSoft application are designed to make it easier for users to fill in the information. Harrison explains: "How do you pass information from the system?" You need different designs, data marts, and Kimball patterns to solve this problem, complete redesign of the data, and "standardize" some of the data so that they adapt to faster reporting and query requirements. ”

On the other hand, Harrison that the Inmon architecture is responsible for designing and maintaining a centralized enterprise data warehouse within the company, which is also running very well. It is important to understand and follow the Inmon theory, which emphasizes the standardization of data when creating an enterprise Data warehouse. Regardless of whether or not a data mart is served in Kimball mode, Harrison believes that this is the place to look.