Thus, PMaps perform better than REBSI for multi-attribute queries
for the VHCAQS.
Analysis
In general, we conclude the following based on our simulations and analysis over
all 6 query sets.
1. The storage cost of REBSI is higher than that of PMap because separate REBSIs
have to be created for each of the attributes accessed in the frequently used
queries. Due to the trade-off between space and time, the scaling factor (sf)
has a predictable impact on the performance of a REBSI. REBSI performance
increases proportionally to increased space allocation.
2. The position of a property covering a predicate in the pstring significantly
affects the number of pages retrieved for a query accessing that predicate.
The closer the property to the beginning of the pstring, the higher the pfilterl
value, which reduces the PMax value, i.e., the number of pages retrieved
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.
by the PMap in the worst case. This is illustrated in all query set observations.
3. PMap performance is not significantly affected by attribute cardinality or query
selectivity.
4. REBSI retrieval cost increases as the number of attributes accessed in the
query increases, even for a very small cardinality attribute like K2.
Pages:
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383