Moreover, Arktos II delves into the logical optimization of ETL workflows, having
as its uttermost goal the finding of the optimal ETL workflow (Simitsis, Vassiliadis,
& Sellis, 2005). The method proposed reduces the execution cost of an ETL work-
flow, by changing either the total number or the execution order of the processes.
The problem is modeled as a state space search problem, with each state representing
a particular design of the workflow as a graph. The tuning of an ETL workflow
is realized through several algorithms for the optimization of the execution order
of the activities.
Finally, to replenish the aforementioned issues, an ETL tool has prototypically been
implemented with the goal of facilitating the design, the (re)use, and the optimization
of ETL workflows. The general architecture of Arktos II comprises a GUI, an
ETL library, a metadata repository, and an optimizer engine. The GUI facilitates
the design of ETL workflows in both the conceptual and logical level, through a
workflow editor and a template palette. The ETL library contains template code of
built-in functions and maintains a template code of user-defined functions. After its
creation, the ETL workflow is propagated to the optimizer in order to achieve a better
version with respect to the execution time. All the aforementioned components
are communicating with each other through the metadata repository.
Pages:
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262