SEARCH
0-9 A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
Prev | Current Page 372 | Next

Robert Wrembel and Christian Koncilia

"Data Warehouses and Olap: Concepts, Architectures and Solutions"

On the
other hand, PMap retrieval cost remains the same or decreases with multi-attribute
queries, as the additional attributes limit the pstring search by changing
pfilterl and pfilterh values.
5. As the database size decreases, REBSI performance becomes better and the
relative savings of the PMap are reduced. This is intuitive since the number
of tuples directly impacts bit-vector length and the number of blocks to read
one bitmap decreases significantly. Savings for PMaps over REBSI are higher
for the larger database sizes and high cardinality attributes, essentially since
the REBSI performance deteriorates in these cases.
6. We identify two strategies to evaluate a disjunctive query, i.e., a query with
one or more OR operators. Strategy A splits a high level query into two or
more separate queries by rewriting it in disjunctive normal form and processing
each disjunct as a separate query, taking the union of the result set as the
final answer. These disjuncts are processed individually and the results for both
constitute the answer to the query. Strategy B processes a high level query as
a single query using multiple pfilters, without separating the disjuncts. We experiment
with both strategies for all the query sets having disjunctive queries
(VHCAQS, LCAQS, HSQS and MQS) and choose the best strategy in each
case in our performance studies.


Pages:
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384