My first Publication Agile-Data-Warehouse-Design-eBook | Page 24

1
✲ !
H OW TO M ODEL A D ATA W AREHOUSE
Essentially, all models are wrong, but some are useful.
— George E. P. Box
In this first chapter we set out the motivation for adopting an agile approach to
data warehouse design. We start by summarizing the fundamental differences
between data warehouses and online transaction processing (OLTP) databases to
show why they need to be designed using very different data modeling techniques.
We then contrast entity-relationship and dimensional modeling and explain why
dimensional models are optimal for data warehousing/business intelligence
(DW/BI). While doing so we also describe how dimensional modeling enables
incremental design and delivery: key principles of agile software development. Dimensional
Readers who are familiar with the benefits of traditional dimensional modeling
may wish to skip to Data Warehouse Analysis and Design on Page 11 where we begin
the case for agile dimensional modeling. There, we take a step back in the DW/BI
development lifecycle and examine the traditional approaches to data requirements
analysis, and highlight their shortcomings in dealing with ever more complex data
sources and aggressive BI delivery schedules. We then describe how agile data
modeling can significantly improve matters by actively involving business
stakeholders in the analysis and design process. We finish by introducing BEAM ✲
(Business Event Analysis and Modeling): the set of agile techniques for collabora-
tive dimensional modeling described throughout this book. Collaborative
Differences between operational systems and data warehouses
Entity-relationship (ER) modeling vs. dimensional modeling
modeling supports
data warehouse
design
dimensional
modeling
supports agile
data warehouse
analysis and design
Chapter 1 Topics
At a Glance
Data-driven analysis and reporting requirements analysis limitations
Proactive data warehouse design challenges
Introduction to BEAM✲: an agile dimensional modeling method
3