Database Reference
In-Depth Information
so on in order to optimize the marketing strategy, for example, detecting
customers that potentially could increase their orders or sales regions that
are underperforming. To be able to conduct the analysis process, Northwind
decided to implement a data warehouse system.
10.3 Requirements Specification
The requirements specification phase is one of the earliest steps in system
development and thus entails significant problems if it is faulty or incomplete.
Not much attention has been paid to the requirements analysis phase in data
warehouse development, and many data warehouse projects skip this phase;
instead, they concentrate on technical issues such as database modeling or
query performance. As a consequence, many data warehouse projects fail to
meet user needs and do not deliver the expected support for the decision-
making process.
Requirements specification determines, among other things, which data
should be available and how these data should be organized. In this phase,
the queries of interest for the users are also determined. The requirements
specification phase should lead the designer to discover the essential elements
of a multidimensional schema, like the facts and their associated dimensions,
which are required to facilitate future data manipulation and calculations. We
will see that requirements specification for decision support and operational
systems differ significantly from each other. The requirements specification
phase establishes a foundation for all future activities in data warehouse
development; in addition, it has a major impact on the success of data
warehouse projects since it directly affects the technical aspects, as well as
the data warehouse structures and applications.
We present next a general framework for the requirements specification
phase. Although we separate the phases of requirements specification and
conceptual design for readability purposes, these phases often overlap. In
many cases, as soon as initial requirements have been documented, an initial
conceptual schema starts to be sketched. As the requirements become more
complete, so does the conceptual schema. For each one of the three approaches
above, we first give a general description and then explain in more detail the
various steps; finally, we apply each approach to the Northwind case study.
We do not indicate the various iterations that may occur between steps. Our
purpose is to provide a general framework to which details can be added and
that can be tailored to the particularities of a specific data warehouse project.
10.3.1 Analysis-Driven Requirements Specification
In the analysis-driven approach, the driving force for developing the con-
ceptual schema are the analysis needs of users. These requirements express
Search WWH ::




Custom Search