Since the real
performance lies in between, we also compute an average of the two.
3. Simulation.framework: Our experiments study the impact of parameters
such as block size, database size, scaling factor, query selectivity, and attribute
cardinality that affect index performance. For each of the query sets, we create
PMaps and REBSIs and measure their performance in terms of the number
Table 7. Parameter values
Data
Specific
VAR DESCRIPTION TYPE VALUES
d dimensionality
of index integer Number of unique attributes
referenced in the query set
t number of tuples integer 50,000 and 1,000,000
c attribute cardinality integer n, where Kn is the attribute
SREC size of a data record bytes 224
System
Specific
SB block size bytes 2048, 4096 and 8192
ws word size bits 16 and 32
User
Defined sf scaling factor integer
minimum sf (sf_min) depending
upon the corresponding PMap,
up to 10
Indexing in Data Warehouses
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
of index pages retrieved. A PMap is created using heuristic design algorithms
(Darira et al., 2006). As a result of the design heuristics, PMap ranges are
designed to exactly cover the query predicates and hence no excess tuples
are retrieved. We calculate the PMap storage requirements and index pages
retrieved for each of the queries, for this query set.
Pages:
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375