is prohibited.
set per bitmap) becomes low and the storage overhead is significantly increased.
Therefore, compression techniques have to be used that will reduce the efficiency
of the bitmap operations.
An Abstract Plan for Star Query Processing
In this section, we will describe the major processing steps entailed when we want
to answer star queries over a hierarchically clustered fact table.
??? Step 1. Identifying relevant fact table data: The processing begins with the
evaluation of the restrictions on the individual dimension tables, that is, the
evaluation of the local predicates (operations Create_Range in Figure 5(a)).
This step performed on a hierarchically encoded dimension table will result in
a set of h-surrogate values that will be used in order to access the corresponding
fact table data. Due to the hierarchical nature of the h-surrogate, this set
can be represented by a number of h-surrogate intervals called the h-surrogate
specification. Using the notation of Karayannidis et al. (2004), an interval
will have the form 1999.*.* for the restriction on the DATE dimension in our
running example. This denotes that we need to access all values under 1999
at the Month level and all values of each such month at the Day level.
Once the h-surrogate specifications are determined for all dimensions, the
evaluation of the star join follows.
Pages:
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292