Optimal Design of Distributed Databases

Abstract

The physical expansion of enterprises with the current trends in globalization has had a positive influence on the technologies of distributed systems. At the forefront of this technological revolution are distributed database systems. For a distributed database to be at optimal performance and thus provide an efficient service it needs to be designed appropriately. The significance of the perfect design is only emphasized by the multiple dimensions required in generating a design. The purpose of this paper is to suggest an approach to generate optimal designs for such distributed database systems and to develop a prototype to demonstrate the said approach. The approach emphasizes on the accuracy of inputs as it largely determines the quality of the final solution. Hence the extraction of network information, a key dimension, is automated to ensure precision. The global schema is fragmented considering data requirements as well as connectivity of each site. Allocation of fragments is treated as a combinatorial optimization problem and assigned to a memetic algorithm. An estimation of distribution algorithm complements the search effort of this memetic algorithm. Site options for replication server environments are investigated based on a shortest path algorithm. Usability of the system in an object oriented development environment, through conditional object-relational mapping, is also explored. The prototype was developed using an evolutionary prototyping approach. It was evaluated by several experts in the relevant fields of application. The results of which, confirmed the practicality of the suggested approach.