At first, there is a necessity for a framework
that will allow the application of several well-known query optimization techniques
to the optimization of ETL workflows. For example, it is desirable to
push selections all the way to the sources, in order to avoid processing unnecessary
rows. Moreover, it is desirable to determine appropriate techniques
and requirements, so that an ETL transformation (e.g., a filter) can be pushed
before or after another transformation involving a function. Additionally, there
is a need to tackle the problem of homonyms. For example, assume the case
of two attributes with the same name, COST, where the first one has values
in European currency, while the other contains values in American currency.
Clearly, in this case it is not obvious if a transformation that involves the first
attribute (e.g., a transformation that converts the values from American to
European currency) can be pushed before or after another transformation that
involves the second attribute (e.g., a transformation that filters the values over
a certain threshold).
Data Warehouse Refreshment 2
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
??? Software.construction: To conclude the discussion about the life cycle of
the data warehouse presented in Figure 2, one anticipates that the outcome
of the aforementioned analysis should be used for the construction of a software
prototype.
Pages:
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257