This case happens when
the removed dimension table was joined only to obtain access to hierarchical
attributes that where then used only for the grouping operation.
2. It can split the grouping operation of the initial plan into two stages: in the
first stage, a pregrouping of the fact table tuples is performed before invoking
them in residual join operations. The grouping is performed using the h-surrogates
contained in the fact table (the foreign keys). In the second stage, which
is executed after performing the required residual joins, the tuples are once
again grouped to obtain the required result. This second grouping operation
Advanced Ad Hoc Star Query Processing
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
uses some of the attributes that where acquired with the residual join operations
performed. This transformation reduces drastically the number of tuples
involved in the residual join operations and therefore it reduces significantly
the cost of the query evaluation plan.
3. It can push join operations above the grouping operation by carefully modifying
the grouping attributes and the join condition. The result of this transformation
is similar to the previous one: it reduces the number of tuples involved in the
affected residual join operations and therefore it improves the overall cost of
the evaluation plan.
Pages:
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305