My first Publication Agile-Data-Warehouse-Design-eBook | Page 160

Modeling Star Schemas
139
Running a Model Review
With the aid of the annotated tables and the event matrix estimates you are ready
for the model review (Figure 5-5) with stakeholders. The purpose of this meeting is
to make stakeholders aware of what could be achieved in the next release, based on
the currently available data sources and task estimates. If the data sources are in
good shape and the business priorities have not changed then it should be a short
meeting. If not, you need to concentrate on the serious data issues and large task
estimates that need the stakeholders’ attention most.
Review the
annotated model
with stakeholders to
make them aware of
what is possible in
the next release
Figure 5-5
Model review
Start with the most severe issues (see Table 5-2) and work your way down by
reviewing any major missing sources first. Completely missing event or dimension
sources—strikethrough tables like Figure 5-3—may cause a serious rethink of
priorities. Missing individual details are generally less disruptive, but some may be
indispensable. Problems with conflicting conformed dimension sources are highly
significant as they can have a knock on effect for future iterations and have the
potential to build up the greatest technical debt.
S EVERITY
I SSUE
O UTCOME
1 Missing conformed dimension Stop
1 Missing event Stop
3 Missing or incorrect business key Stop
3 Conflicting data for conformed dimension Stop
5 Event granularity is different Stop/Pause
6 Missing non-conformed dimension Pause
6 Missing (or poorly populated) event detail Pause
8 Missing mandatory values Pause
8 Incorrect hierarchical relationship Pause
10 Missing (or poorly populated) dimensional attribute Go
10 Mismatched detail and attribute values Go
12 Additional event details or dimensional attributes Go
Stop : a major rethink or reprioritization is necessary.
Pause : provide feedback before you develop the physical database.
Go
: proceed (with caution) but still provide feedback on the gaps in
the model or the additional BI opportunities that may exist.
Concentrate on
severe data issues:
missing sources and
conflicting sources
for conformed
dimensions
Table 5-2
Data source issues
ranked by severity
(1=highest,
12=lowest)