Performance Evaluation using Various Models in Distributed Component based Systems

Over the years, several models were proposed to analyze the performance of distributed component based system with the view of improving system’s performance. Traditional methods of prediction such as Petri nets and queuing networks exploit the benefits of component engineering paradigm, such as division of work and reuse. We have surveyed different approaches of performance prediction which have attained widespread industrial use. Each approach has different goals and context, at which phase of life cycle process which approach is appropriate is described, there benefits and drawbacks are also given. Based on the analysis we have found that the models are machine centric e.g. throughput, responsiveness, no. of processes CPU has to execute within limited processor speed. None of the model draws parameters from the contributions of the client organization and end users. We have proposed a solution in this paper for taking into consideration end users perspective. Software is developed in order to satisfy requirements of the end users. Therefore, involving users in evaluating performance should not be underestimated