My first Publication Agile-Data-Warehouse-Design-eBook | Page 37

16
Chapter 1
Agile Data Warehouse Design
Traditional data
warehousing follows
a near-serial or
waterfall approach
to design and
development
Traditional data warehousing projects follow some variant of waterfall develop-
ment as summarized on the Figure 1-7 timeline. The shape of this timeline and the
term ‘waterfall’ might suggest that its ‘all downhill’ after enough detailed require-
ments have been gathered to complete the ‘Big Design Up Front’ (BDUF). Unfor-
tunately for DW/BI, this approach relies on a preternatural ability to exhaustively
capture requirements upfront. It also postpones all data access and the hoped for
BI value it brings until the (bitter) end of the waterfall (or rainbow!). For these
reasons pure waterfall (analyze only once, design only once, develop only once,
etc.) DW/BI development, whether by design or practice, is rare.
Figure 1-7
Waterfall DW
development
timeline
Dimensional
modeling enables
incremental
development
Agile data
warehousing is
highly iterative and
collaborative
Figure 1-8
Agile DW
development
timeline
Dimensional modeling can help reduce the risks of pure waterfall by allowing
developers to release early incremental BI functionality one star schema at a time,
get feedback and make adjustments. But even dimensional modeling, like most
other forms of data modeling, takes a (near) serial approach to analysis and design
(with ‘Big Requirements Up Front’ (BRUF) preceding BDUF data modeling) that
is subject to the inherent limitations and initial delays described already.
Agile data warehousing seeks to further reduce the risks associated with upfront
analysis and provide even more timely BI value by taking a highly iterative, incre-
mental and collaborative approach to all aspects of DW design and development as
shown on the Figure 1-8 timeline.