Technical DW/BI System Architecture（技术系统架构）
Technical DW/BI System Architecture（技术系统架构）
The Kimball technical system architecture separates the data and processes comprising the DW/BI system into the backroom extract, transformation and load (ETL) environment and the front room presentation area, as illustrated in the following diagram.
Figure 1: Kimball technical system architecture diagram.
The Kimball technical system architecture focuses on the following components:
- Backroom ETL system: The Kimball Group has identified 34 subsystems in the ETL process flow, grouped into four major operations: extracting the data from the sources, performing cleansing and conforming transformations, delivering it to the presentation server, and managing the ETL process and back room environment.
- Front room presentation area: The Kimball Architecture presumes the data utilized by the BI applications is dimensionally-structured, organized by business process, atomically-grained (complemented by aggregated summaries for performance tuning), and tied together by the enterprise data warehouse bus architecture, as described earlier on this page.
- Front room BI applications: The front room is the public face of the DW/BI system; it’s what business users see and work with day-to-day. There’s a broad range of BI applications supported by BI management services in the front room, including ad hoc queries, standardized reports, dashboards and scorecards, and more powerful analytic or mining/modeling applications.
- Metadata: Metadata is all the information that defines and describes the structures, operations, and contents of the DW/BI system. Technical metadata defines the objects and processes which comprise the DW/BI system. Business metadata describes the data warehouse contents in user terms, including what data is available, where did it come from, what does it mean, and how does it relate to other data. Finally, process metadata describes the warehouse’s operational results.
Some organizations adopt an alternative data warehouse architecture that includes a third normal form (3NF) relational data warehouse. This hub-and-spoke architecture, often called the Corporate Information Factory (CIF), includes a data acquisition ETL process to gather, clean and integrate data similar to the backroom ETL system described above. With the CIF, atomic data is loaded into third normal form structures, typically called the enterprise data warehouse (EDW). Another ETL data delivery process then populates downstream reporting and analytic environments supporting the business users; these environments are typically structured dimensionally.
A modification to the Kimball Architecture, sometimes referred to as a hybrid architecture, leverages an existing 3NF data warehouse as the source of clean, integrated data to feed the front room presentation area described above.
The following articles provide an overview of the Kimball Architecture. Full coverage is available in The Data Warehouse Lifecycle Toolkit, Second Edition.
- “Two Powerful Ideas”, Intelligent Enterprise, September 2002
- “Data Warehouse Dining Experience”, Intelligent Enterprise, January 2004
- “Differences of Opinion”, Intelligent Enterprise, March 2004