A Multistep Approach for Restructuring and Mapping Distributed Object Oriented Software onto a Multiprocessor System

Abstract

A large number of Distributed Object Oriented (DOO) systems have been developed for solving complex problems in various scientific fields. In a distributed object-oriented application, classes can run on a separate computer within a network system. So, they should be distributed efficiently among different nodes. However, the initial design of the DOO application does not necessarily have the best class distribution. In such a case, the DOO software may need to be restructured. The challenge resides in the complexity of interactions between objects. In this paper, we present an approach for efficiently restructuring the DOO software classes in order to be mapped onto target distributed architecture. The proposed methodology consists of a set of consecutive steps. In the first step we are using the Distributed Object-Oriented performance (DOOP) model as a technique for the assessment of relationships between system classes. Next, a recursive graph clustering technique is utilized to partition the OO system into subsystems that have low coupling and are more suitable for distribution. Then, the mapping process is applied to map the generated partitions to the nodes of the target architecture. An experimental case study is then studied to illustrate the results after applying each step of the proposed approach.