Object clustering techniques for client/server and distributed software development

While object-oriented paradigm is known to be effective for developing web-based software, most of the current object-oriented methodologies including OMT and UML do not provide effective methods and instructions for modeling client/server or distributed applications. We present new effective object clustering formalism and algorithms for client/server and distributed application development. The coupling degree among objects is often the key factor in grouping related objects into a cluster. If objects with large coupling degree are identified and clustered together, the cohesion in the cluster is increased and so network traffics among the clusters are largely reduced, yielding a high overall system performance. The paper presents the metrics for measuring object coupling and shows systematic algorithms for object clustering based on the proposed metrics. In addition, a case study is given to assess the proposed formalism and methods.