Well distributed data can dramatically improve the efficiency and effectiveness of the use of distributed database systems to satisfy geographically dispersed data processing demands. Among several issues related to distribution design in distributed databases, data allocation design is of major importance. Choices of a fragmentation strategy and location of database files are two critical decisions. Thus far, solutions of these design problems, although interdependent, have been attempted separately. Solving both design problems simultaneously in a real design setting is not a trivial task. By formulating typical data allocation design problems, we can analyze the solution space and analytical properties of optimal data allocation design. Based on this, we suggest that clustering data elements into uniform fragments and then allocating these fragments is equivalent to solving the data allocation design as a whole. Such analytical examination of the data allocation design problem has not been attempted by other researchers, but it is essential to provide the theoretical foundation for solving the fragmentation design and fragment allocation design problem. We then extended the research by studying the effect on design issues of such characteristics of distributed processing as database access patterns, network scope, and design objectives. We also propose a generic taxonomy of data allocation design models. We further advance data allocation design skills in the following two directions. The first of these involves developing a design method that guarantees the minimum number of fragments to be considered as units of allocation. This improves upon existing fragment allocation methodologies, which are based on the assumed units of allocation. The second direction involves enhancements in modeling and solution procedures that allow efficient fragment allocation design. Concentration is on information processing environments, which have received little attention in the research literature. We first studied databases connected on local area networks under weak locality of reference. The model proposed is validated by simulation study. We then explored the multiple design objective optimization phase, which involves searching for models where several design objectives are in conflict. We addressed three important design objectives including response time, operating cost and data availability. In conclusion, we submit that the methodology proposed is likely to provide a better understanding of data allocation design problems, the solutions for which are expected to continue providing key design tools as advancing data communication techniques evolve.
| Identifer | oai:union.ndltd.org:arizona.edu/oai:arizona.openrepository.com:10150/185435 |
| Date | January 1991 |
| Creators | Lee, Heeseok. |
| Contributors | Liu Sheng, Olivia R., Law, Averill M., Sanchez, Susan M. |
| Publisher | The University of Arizona. |
| Source Sets | University of Arizona |
| Language | English |
| Detected Language | English |
| Type | text, Dissertation-Reproduction (electronic) |
| Rights | Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author. |
Page generated in 0.0019 seconds