Abstract
We summarize the problem tackled here in the following way: Given a modern database application environment, how can we identify and select the database structure that provides robust performance across changing query patterns and arrival rate conditions? We demonstrate the importance of investigating the underlying relationships and then utilize this information in formulating robust structures. Our work is pre-theory in the philosophy of science sense. That is, the careful identification and observation of relationships will subsequently be utilized in formulating a testable theory of the development of robust database structures under dynamic query patterns and arrival rates. Our first step in providing a database design or "structure selection" method is to determine potential good performers among different database structures. These potential good performers are selected and analyzed across arrays of query patterns. The next step is to identify database structures that are robust structures, that is good performers across the different types of query patterns and arrival rate levels. The presentation includes illustrations of the determination of actual query pattern processing times and the use of these times within a queuing analysis. In fact, for the database layout analyzed, application of our methods demonstrates the existence of such robust database structures.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 435-447 |
| Number of pages | 13 |
| Journal | Decision Support Systems |
| Volume | 37 |
| Issue number | 3 |
| DOIs | |
| State | Published - Jun 2004 |
Bibliographical note
Funding Information:Dr. James R. Marsden came to UConn in 1993 as Professor and Head, Department of Operations and Information Management, School of Business Administration, University of Connecticut. Dr. Marsden was part of a three-person concept development team that initiated and oversaw the development of the Connecticut Information Technology Institute and is currently serving as its Executive Director. He developed and implemented the Treibick Electronic Commerce Initiative that is funded through a generous gift provided by Richard Treibick and the Treibick Family Foundation. He was a member of the edgelab development team and currently serves on the edgelab Steering Committee which selects and resources projects and oversees operations. Dr. Marsden is a tow-time winner of the Chancellor's Award for IT Excellence and a co-winner of the Team Connecticut Program Award from the Office of Economic Development. He has a lengthy research publication record in market innovation and analyses, economics of information, artificial intelligence, and production theory. His research work has appeared in Management Science ; IEEE Transactions on Systems, Man, and Cybernetics ; American Economic Review ; Journal of Economic Theory ; Journal of Political Economy ; Computer Integrated Manufacturing Systems ; Decision Support Systems ; Journal of Management Information Systems , and numerous other academic journals. Professor Marsden received his AB from the University of Illinois and his MS and PhD from Purdue University. Having completed his J.D. while at the University of Kentucky, Jim has been admitted to both the Kentucky and Connecticut Bar.
Funding Information:
The authors are grateful for support from the Shenkman Family Chair, the Gladstein Professorship, the Gladstein MIS Research Lab and the Treibick Electronic Commerce Initiative without which this work could not have been completed.
Keywords
- Database design
- Query pattern
- Query processing
- Queueing