, 2006).
The analyses also indicate that the worst-case behavior is for attributes following
.E+0
.E+08
.E+0
Base data EE- 000-lit EE- 000-
comp
RE- 00-lit RE- 00-
comp
Base data and indices for high-energy physics data set
Size [bytes]
Note: For an explanation of the legend, see Figure 6.
Figure 7. Size of base data compared with bitmap indices
Bitmap Indices for Data Warehouses
Copyright ?© 2007, Idea Group Inc. Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
a uniform random distribution. Figure 8 plots the query response time against the
number of hits for a set of queries on two attributes with different attribute cardinalities.
The data values for the two attributes are randomly distributed in the range of
[0;100] and [0; 10,000] respectively. We see that in both cases the timing measurements
follow straight lines, which is theoretically optimal.
In the remainder of this section we present more timing measurements to compare
the query response time of equality-encoded and range-encoded bitmap indices.
All indices are compressed with WAH compression. Since the results for the two
datasets are similar, we only report on the measurements based on the larger and
thus more challenging combustion dataset. We use the projection index as the base
line for all the comparisons.
Pages:
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341