Author

Hababeh

Date of Award

12-1996

Degree Name

Master of Science

Department

Computer Science

First Advisor

Dr. Dalia Motzkin

Second Advisor

Dr. Dionysios Kountanis

Third Advisor

Dr. Mark Kerstetter

Access Setting

Masters Thesis-Open Access

Abstract

This thesis presents a strategy for the fragmentation and allocation of data in a distributed relational database. The strategy introduces a way to fragment and allocate data to accomplish the benefits of allocating each fragment independently for each site. Some fragments may be allocated to multiple sites in order to achieve some constraints such as availability, consistency, and integrity. The strategy is appropriate for different types of computers at various sites connected to a network, with different costs of retrieval, update, and communication.

To increase the system 1/0 performance, sites having similar communication costs are grouped into clusters to which fragments would allocated. The strategy describes a data allocation techniques based on a heuristic algorithm which uses the allocation benefit as a difference between cost of not allocating data to clusters and sites, and the cost of allocating data to the same clusters and sites. The proposed overall strategy presents near optimal data allocation in the system and is determined in polynomial time.

Share

COinS