The fragmentation problem in distributed databases

Abstract

Data Fragmentation is a heuristic problem in the design of a distributed database. Its purpose is to maximize the locality of reference, minimize data access at remote site, and to decrease the number of disk accesses in the system. In this thesis, we review the design a homogeneous distributed database. As the design is a heuristic problem, we also review a number of algorithms suggested as solutions to the three types of fragmentation: vertical, horizontal or mixed. Next, we propose that a previously implemented routine entitled "a transaction-based vertical partitioning algorithm" can be implemented when fragmenting the database horizontally. Finally, a simulation of the proposed technique is presented.