A Method for Data Warehouse Design - Data Warehouse Systems: Design and Implementation - page 408

View materialization is a key feature for achieving good query perfor-
mance in OLAP. Thus, during physical design, we must define which are
the most common queries our system will support and, based on this study,
define which are the materialized views that we need. Recall that SQL Server
supports materialized views indirectly, through the feature called indexed
views, basically a unique clustered index defined over a view. Since view
materialization was studied in Chap. 7 , we do not extend further in this topic
here.
10.7 Characterization of the Various Approaches
In this section, we summarize the three approaches to data warehouse
development. We discuss the many aspects that must be considered before
choosing one of those approaches for a specific data warehouse project.
10.7.1 Analysis-Driven Approach
The analysis-driven approach requires the intensive participation of users
from different organizational levels. In particular, the support of executive-
level users is important in order to define business goals and needs. The
identification of key users for requirements specification is a crucial task. It
is necessary to consider several aspects:
￿ Users should be aware of the overall business goals to avoid situations
where the requirements represent the personal perceptions of users accord-
ing to their role in the organization or business unit.
￿ Users who would dominate the requirements specification process should
be avoided or tempered in order to ensure that the information needs of
different users will be considered.
￿ Users must be available and agree to participate during the whole process
of requirements gathering and conceptual design.
￿ Users must have an idea of what a data warehouse system and an OLAP
system can offer. If this is not the case, they should be instructed by means
of explanations, demonstrations, or prototypes.
The development team requires highly qualified professionals. For example,
a project manager should have very strong moderation and leadership skills.
A good knowledge of information-gathering techniques and business process
modeling is also required. It is important that data warehouse designers
should be able to communicate with and to understand nonexpert users
in order to obtain the required information and, later on, to present and
describe the proposed multidimensional schema to them. This helps to avoid
the situation where users describe the requirements for the data warehouse