Agile Data Warehousing for the Enterprise

Chapter 12

Traditional Data Modeling Paradigms and Their Discontents

The standard normal forms and conformed dimensions that undergird most data warehouses built until before agile data engineering became available unfortunately result in applications that are very expensive to modify once data is loaded into their integration layers. Four change cases illustrate how expensive the changes to traditionally structured integration layers can be. Splitting out attributes into tables of their own and generalizing customers and employees to a “party model” can require several thousand hours of work when data conversion requirements are included. For data warehouses that load data directly to star schemas, changing the variables that trigger updates ...