Data mart or data warehouse?
This is a question that arises much passion, depending upon personal experience.
In this context the question is whether to take a departmental or corporate
approach (and we are not referring to the Kimball vs Inmon
debate). Data marts tend to be departmental, and data warehouses
corporate.
First the cynical view.
-
A corporate data warehouse is a monolithic, IT driven, expensive solution that
requires an elephant’s pregnancy to deliver information that is not used in the
most complicated fashion.
-
Data marts are the best way to deliver multiple inconsistent repositories with
duplicated data and multiple versions of the truth where each department fights
for ‘their’ system.
Most IT professionals tend to avoid departmental approaches where a corporate
one can be employed. It fits our sense of standards and orderliness, and we’ve
all seen the unsupportable and costly mess that ad hoc delivery can cause. We
tend to associate departmentally driven solutions with more ad hoc focused
delivery.
Business Intelligence should provide information to those business users smart
enough and willing enough to modify and improve their behaviour. It is a true
adage that without enthusiastic users a BI solution will invariably fail. So,
if your organisation has the required BI zeal at a departmental but not at a
corporate level, then ignore your instincts to push for a corporate solution.
You will be benefiting someone so it will be worthwhile. Obviously, if you can
get the required corporate support, then go for it.
Read next »