As research challenges in this area, we mention
the following issues:
?‚? A formal description of ETL processes with particular emphasis on an
algebra (for optimization purposes) and a formal declarative language.
?‚? The optimization of ETL processes on logical and physical levels. A
challenge will be either the optimization of the whole ETL process or of
any individual transformation. Parallel processing of ETL processes is
of particular importance.
?‚? The propagation of changes back to the sources. Potential quality problems
observed at the end-user level can lead to clean data being propagated
back to the sources, in order to avoid the repetition of several tasks in
future application of the ETL process. Clearly, this idea has already been
mentioned in the literature as ???backflow of cleaned data??? (Rahm & Hai
Do, 2000), but the problem is not solved yet.
?‚? The provision of standard-based metadata for ETL processes. There does
not exist common model for the metadata of ETL processes. CWM is
not sufficient for this purpose and it is too complicated for real-world
applications.
?‚? The integration of ETL with XML adapters, EAI (Enterprise Application
Integration) tools (e.g., MQ-Series), and data quality tools.
?‚? The extension of the ETL mechanisms for nontraditional data, like XML/
HTML, spatial, and biomedical data.
Pages:
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264