In our solution, all of this
is done in a ???hub??? with the intent to minimize and parallelize I/O, saving elapsed
time and resource usage as much as possible. However, before dealing with the
issue in the whole (the hub), let us look at the several specific aspects: acquisition
scenario and loading techniques.
Acquisition.Scenario
We can acquire data from files, directly from DBMS, or listening to incoming messages,
but this last eventuality is more typical in mediation systems than in ETL.
In the scholastic ETL examples, there is always a source db, a target db, and in
the middle, the ETL processes. In our experience, we never had the opportunity to
acquire the data we needed directly from the source DBMS. It is politically hard
obtaining access to a system owned by another division, group or department, and
this is the typical organizational scenario in a DW project. It should not be very
reasonable for a system manager to open a system to others and allow the installation
of foreign agents for many understandable reasons. A better and more usual
way is to define (the owner of data defines) an interface in order to decouple the
source from the target system. The simplest method to do that is the use of flat files.
Direct extraction from source DBMS often is not possible; it does not permit decoupling
systems, DBMS may be different (e.
Pages:
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214