| G. C. Steindel and H. G. Madison. A Benchmark Comparison of DB2 and the DBC/1012. In CMG '87, International Conference on Management and Performance Evaluation of Computer Systems, pages 360--369, Orlando, FL, 1987. The Computer Measurement Group, Inc. |
....possible partial indexing strategies described in this paper, we call our strategy generalized partial (GP) indexing. A significant contribution of this paper is to demonstrate the importance of using statistics on data distributions. The data distributions in real databases are seldom uniform [2, 8], and it is often more useful to index those tuples with values in a sparse region of the column domain. This is demonstrated graphically in Figure 3. Further, the distribution of the query values (the actual constants in the predicates of the queries) is often quite different from the ....
....The relation holds 100,000 records, resulting in an 80MB database. There are 10 columns to be indexed, each having 10,000 distinct values. The data distribution in each column is a generalized Zipfian distribution [5] the Zipfian distribution models a significant amount of real skew data[2, 8]) with Zipf parameter = 1:03. For the chosen database configuration parameters, this corresponds to an 80 20 distribution[5] The indexed columns are integer valued (4 bytes) and along with the RID, each index entry has a size of 12 bytes. With an index fill factor of 70 [10] this ....
G. C. Steindel and H. G. Madison. A Benchmark Comparison of DB2 and the DBC/1012. In CMG '87, International Conference on Management and Performance Evaluation of Computer Systems, pages 360--369, Orlando, FL, 1987. The Computer Measurement Group, Inc.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC