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 321 | Next

Robert Wrembel and Christian Koncilia

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


Since WAH compressed bitmaps are larger in size than BBC compressed bitmaps,
we would expect that WAH require more I/O time to read compressed bitmaps. For
many database operations, the CPU time is negligible compared with the I/O time.
It turns out that this is not the case when answering queries with compressed bitmap
indices. In a performance experiment Stockinger et al. (2002) compared WAH
compressed indices with two independent implementations of BBC compressed
indices, one based on Johnson??™s (1999) code and the other by Wu et al. (2002). The
results showed that the total query response time was smaller with WAH compressed
bitmap indices than with BBC compressed bitmaps, even on a relatively slow disk
system that can only sustain 5 MB/s for reading files from disk. On faster disk systems,
the performance advantage of WAH compressed bitmap indices is even more
pronounced. Using WAH could be 10 times faster than using BBC.
Bitmap. Index.Tuning
Unless one uses binary encoding, it is important to compress the bitmap indices. To
build an efficient compressed bitmap index, the three main parameters to consider
are: (1) encoding, (2) number of bins, and (3) binning strategy. In the following we
present a rule-of-thumb for choosing these three parameters.
The optimal bitmap encoding technique depends on the kind of queries that are
evaluated.


Pages:
309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333