After reviewing these strategies, we give a more in-depth
discussion on how the word-aligned-hybrid (WAH) bitmap compression technique
reduces the bitmap index sizes. We will also present some timing results to demonstrate
the effectiveness of the WAH compressed bitmap indices for two different
application datasets. Our performance evaluation is deliberately based on datasets
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.
with high-cardinality attributes, since for low-cardinality attributes the performance
advantage of bitmap indices is well known. We conclude with a short review of
bitmap indices available in commercial DBMS products and discuss how to make
bitmap indices better supported in these commercial products.
Background
By far the most commonly used indexing method is the B-Tree (Comer, 1979).
Almost every database product has a version thereof since it is very effective for
online transaction processing (OLTP). This type of tree-based indexing method has
nearly the same operational complexities for searching and updating the indices.
This parity is important for OLTP because searching and updating are performed
with nearly the same frequencies. However, for most data warehousing applications
such as online analytical processing (OLAP), the searching operations are typically
performed with a much higher frequency than that of updating operations (Chaudhuri
& Dayal, 1997; Chaudhuri, Dayal, & Ganti, 2001).
Pages:
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315