My first Publication Agile-Data-Warehouse-Design-eBook | Page 149

128
Chapter 4
Summary
“Just barely good enough” dimensional modeling can lead to the early and frequent deployment
of data marts that answer current departmental reporting requirements, but it also stores up
technical debt, in the form of incompatible data silos that cannot support cross-process analysis
and enterprise level BI. Due to the large data volumes associated with DW/BI, repaying this
debt can be ruinous.
To avoid silo data marts and reduce technical debt, agile DW designers need to model ahead of
the current development sprints and release plans, just enough to identify and define conformed
dimensions. These reusable components of a dimensional model enable drill-across reporting by
providing the consistent row headers and filters needed to combine and compare measures
from multiple business processes. A well documented, well publicized and well maintained set
of conformed dimensions form a data warehouse bus architecture that supports the incremental
development of truly agile data marts.
Conformed dimensions are single dimensions tables or synchronized copies shared by multiple
star schemas. They can also be swappable [SD] subsets or rollups [RU], derived from a base
dimension, conformed at the attribute level with identical business meaning and contents.
Generalized conformed dimensions that play multiple roles in the same or different events are
referred to as role-playing [RP] dimensions.
Agile dimensional modelers define conformed dimensions by modeling with examples, with
business stakeholders. BEAM ✲ example data stories highlight the value of conformance to the
very people who can make it happen politically. Examples can quickly expose the inconsistent
business terms that would hinder conformance.
The event matrix is a modeling and planning tool that documents the relationship between
events and dimensions. It acts as a storyboard for an entire data warehouse design showing just
enough detail to help identify the most valuable conformed dimensions and prioritize their
development.
Listing events in time/value sequence on an event matrix helps you discover missing events by
highlighting large time gaps or value jumps in process workflows. It also helps you identify
strict chronological process sequences: candidate evolving events, that combine all the milestone
events of a business process, to support end-to-end process performance measurement.
When modeling new events, abbreviated examples allow you to quickly tell stories by reusing
conformed examples where applicable. Unlike codes, abbreviations help to keep stories brief
and readable for stakeholders. They also support the validation, reuse and enhancement of
conformed dimensions.