Extraction, Transformation, and Loading Processes
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
Physical Database Design
Many important works related to the database modeling and design for DW are available
in Kimball et al. (1998); in this chapter, we focus on physical database design
and correlated techniques because this aspect is crucial for the whole ETL process;
a powerful transformation engine working on a not-well structured database may
be a cause of problems for the entire ETL process. In this chapter, we will describe
some important issues in physical database design, in particular, the relevance of
partitioning (range, hash, composite) for managing history and for performance issues
that must be taken into account in building a DW/ETL project.
Partitioning is a very important issue in a database design, in OLTP and DW for
scalability reasons and, in the case of DW, for managing history. Partitioning is the
ability to divide/organize data in separate subsets, for example, the ability to split
a table into several ones still viewing it as a whole. When we have more than one
database in different locations, we have done some kind of partitioning, when we use
different tables to contain the same types of data. In all these cases, there is partitioning,
and it does not matter if we use or do not use the partitioning constructions.
Pages:
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196