My first Publication Agile-Data-Warehouse-Design-eBook | Page 161

140
Chapter 5
Revise the model
with the help of
the stakeholders
As you go through each table or column issue, update the model (the individual
tables and the matrix) with the stakeholders assistance. Ask them to help you to
decide:
Should we include, exclude, add or adjust this item?
Which of these conflicting sources should we choose?
!
If stakeholders want
to reprioritize
events, revise the
matrix accordingly
You should finish the review by asking the stakeholders if they want to reprioritize
events in light of the data issues and task estimates—bearing in mind that the task
estimates may also need to be adjusted based on the changes they have just agreed
to. If the stakeholders do want to alter their priorities, revise the matrix by replay-
ing the event rating game described in Chapter 4.
Sprint Planning
Use the revised
model, estimates
and priorities to
define the sprint
backlog
Following the model review, you hold a sprint planning meeting (Figure 5-6)
where the DW/BI team will revise their estimates based on the model amendments
and the product owner will decide on the data items that will make their way onto
the sprint backlog: the list of data and user stories (tables and BI re-
ports/dashboards) to be implemented in the next sprint. To help the team revise
their estimates you may need to draw some quick star schemas. It is at this point
you would introduce some of the design patterns described later in this book.
Figure 5-6
Sprint planning
meeting
Dimensions that have already been implemented should be given an estimate of
zero. The estimate for all non-viable or low priority tables (that have not been
profiled) should be left blank. The estimate for degenerate dimensions should also
be blank—their development effort is included in fact table estimates. The total
estimate for two star schemas that share conformed dimensions should not
double count the conformed dimension estimates—the conformed dimension
estimates should be high enough individually to include all the tasks involved in
merging and conforming multiple attribute sources.