Database Reference
In-Depth Information
It reduces the user involvement required to start the project.
It facilitates a fast and straightforward development process, provided that
well-structured and normalized operational systems exist.
It allows automatic or semiautomatic techniques to be applied if the
operational databases are represented using the entity-relationship model
or normalized relational tables.
However, it is important to consider the following disadvantages before
choosing this approach:
Only business needs reflected in the underlying source data models can be
captured.
The system may not meet users' expectations since the company's goals
and the users' requirements are not reflected at all.
The method may not be applied when the logical schemas of the underlying
operational systems are hard to understand or the data sources reside on
legacy systems.
Since it relies on existing data, this approach cannot be used to address
long-term strategic goals.
The inclusion of hierarchies may be dicult since they may be hidden in
various structures, for example, in generalization relationships.
It is dicult to motivate end users to work with large schemas developed
for and by specialists.
The derivation process can be dicult without knowledge of the users'
needs since, for instance, the same data can be considered as a measure or
as a dimension attribute.
10.7.3 Analysis/Source-Driven Approach
As this approach combines the analysis-driven and source-driven approaches,
the recommendations regarding users and the development team given above
should also be considered here. The analysis/source-driven approach has
several important advantages:
It generates a feasible solution, supported by the existing data sources,
which better reflects the users' goals.
It alerts about missing data in the operational databases that are required
to support the decision-making process.
If the source systems offer more information than what the business users
initially demand, the analysis can be expanded to include new aspects not
yet considered.
However, this approach has the following disadvantages:
The development process is complicated since two schemas are required,
one obtained from the definition of the analysis requirements and another
derived from the underlying source systems.
Search WWH ::
Custom Search