..
...
(a)
(b)
Advanced Ad Hoc Star Query Processing 147
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
Processing Star Queries over CUBE File Organized
Fact Tables
The CUBE File (Karayannidis et al., 2004) is a multidimensional data structure for
storing the most detailed data of a fact table. Thus it could be exploited as an alternative
primary organization to heap files for fact tables. It provides fast indexing
on data, when these are accessed via restrictions on the hierarchies. Moreover, it
physically clusters data w.r.t. dimension hierarchies (i.e., hierarchical clustering),
which reduces significantly the I/O cost for star query evaluation.
The CUBE File partitions the multilevel-multidimensional data space of an OLAP
cube in disjoint subspaces, called chunks, which are formed by all hierarchy value
combinations per hierarchy-level. This process is called hierarchical chunking (Figure
6(a)) and results in a chunk-tree representation of the cube (Figure 6(b)). Note
that prior to applying hierarchical chunking all hierarchies have to be normalized
to the same length with the insertion of pseudo-levels to the shorter ones. The main
advantage of hierarchical chunking is that it results in a structure that is highly adaptive
to the cube??™s inherent extreme sparseness.
Pages:
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295