Strategy A performs better in general, except
for the queries HC11 and HC12 in the VHCAQS. These queries have multiple
conditions on the attribute KSEQ and Strategy B performs much better for
them.
Conclusion
In this section we present techniques and enhancements that emerge from analyzing
query processing performance of PMaps. Further study of these techniques
could result in improved PMaps and guidelines for using different PMap creation
techniques and query processing strategies in different scenarios. Based on all
observations and analyses, we give the following general guidelines for the use
of PMaps and applications where they may be useful and provide extra savings in
query processing.
200 Davis & Gupta
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of
Idea Group Inc. is prohibited.
??? From our analysis, we learn that PMap performance is not affected by attribute
cardinality; however the property position in the pstring is a significant
factor. Since performance for most other indexing techniques deteriorates
for high cardinality attributes, we can achieve significant savings for these
by creating PMaps on high and very high cardinality attributes. The property
ordering should be in decreasing order of savings desired for the attributes.
Thus, the attributes for which we desire maximum savings have properties at
the beginning of the pstring.
Pages:
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385