Enterprise Data Warehouse (EDW)

Inmon’s approach started by setting up data model of all enterprise data to implement enterprise data warehouse (Connelly & Begg, 2014). Departmental databases that is, data mart are fed data by enterprise data warehouse (EDW). EDW also supports specialized decision support applications like customer relationship management (CRM). Foundation of Inmon’s corporate information factory (CIF) is traditional database methods and techniques to setup EDW.

On the other hand, Kimball’s method employs new methods and technologies to setup EDW. It starts with the creation of document called data warehouse bus matrix which encompass all key business processes and directions to perform analysis (Connelly & Begg, 2014). Matrix is meant to setup data mart for a particular group of enterprise by using dimensionality modeling based on star schema which is highly denormalized. Integration of several data marts ultimately lead to development of EDW.

Inmon’s approach works well when the development of complete enterprise’s data is critical instead of particular group’s data mart, which is why it works best for an enterprise afford to take on a large project lasting more than a year for return of interest (ROI). On the contrary, Kimball’s methodology suits requirement of particular group within an enterprise on a short duration. From business dimensional lifecycle perspective;

1. Inmon’s apprach offers comprehensive and consistent view of enterprise data whereas Kimball’s approach meant for scaled down projects requiring quick turnaround.

2. Pitfall of Inmon’s is that large project often fail to finish in allotted time period and budget whereas Kimball’s methodology works best to meet time and budget goals but struggles to get consistent and comprehensive view after all department data marts are brought together.

Extraction, transformation, and loading (ETL) under Inmon’s methodology happen to physically normalized and non-redundant departments of large enterprises like sales, purchasing, inventory etc. thus making data loading less complex but hard because of so many tables and joins (Rangarajan, 2016). Thus more ETL work is needed for Inmon’s CIF because the data mart of individual departments are built from data warehouse. Kimball’s methodology require data loading into dimensional model, which is not normalized unlike Inmon’s CIF and setup in star schema. In start schema, fact table are of given subject area is surrounded by dimensions which is why drilling up and down is easier for user without joining tables.

George (2012) defines Inmon’s approach as top-down while Kimball’s as bottom-up. In the staging phase of Inmon’s CIF, ETL is performed resulting in the creation of normalized EDW broken into data marts for analysis, reporting and mining. On the other hand, Kimball’s ETL stage will generate start schema consisting of denormalized dimension and fact table with some redundancy to provide data access for analysis, reporting and mining.

Reference

Connolly, T. & Begg, C. (2014). Database Systems: a practical approach to design, implementation, and management (6th Ed.). Upper Saddle River, NJ: Pearson.

George, S. (2012). Inmon or Kimball: which approach is suitable for your data warehouse? Retrieved from http://www.computerweekly.com/tip/Inmon-or-Kimball-Which-approach-is-suitable-for-your-data-warehouse

Rangarajan, S. (2016). Data Warehouse Design – Inmon versus Kimball. Retrieved from http://tdan.com/data-warehouse-design-inmon-versus-kimball/20300