Item – Theses Canada

​
OCLC number
77045612
Link(s) to full text
LAC copy
LAC copy
Author
Li, Chao,1977-
Title
A partial pre-computation of aggregates for MOLAP database.
Degree
M. Sc. -- Simon Fraser University, 2004
Publisher
Ottawa : Library and Archives Canada = Bibliothèque et Archives Canada, [2006]
Description
1 microfiche
Notes
Includes bibliographical references.
Abstract
Partial pre-computation for OLAP (On-Line Analytic Processing) databases has been a popular research topic in recent years. Obviously, a partial pre-computation scheme should be chosen such that the answering time of a particular query workload is optimised. The query workload of an OLAP application is the set of queries users expect. Most of the papers published only deal with the optimisation for the workload of views within the context of ROLAP (Relational OLAP). The partial pre-computation schemes optimised for views have been criticized for lack of support to ad-hoc querying and that a view may be too large a unit for pre-computation for queries that fetch very small answers. In this thesis, we study the problem of partial pre-computation for the workload of point queries, which is suited for MOLAP (Multidimensional OLAP) systems. The point queries in our workload are range queries defined by the dimension hierarchies. Our study shows that without careful coordination between pre-computation and query processing, it would be difficult to realize any significant gain in query performance from the pre-computation for point queries. We present a new organization of pre-computed cells, Pre-computation Cube or PC Cube, upon which we devise a cover-based querying processing method that is efficient and effective. We also design an algorithm for the selection of the PC Cube. Experiments are performed to show (i) the efficiency of the cover-based query processing method, (ii) the effectiveness of the PC Cube and our selection algorithm, and (iii) the performance comparison of variations of our selection algorithms.
ISBN
0494034971
9780494034972