On study of the performance evaluation of the autonomous distributed computing system - Simulation Approach

摘要

In this paper, we describe the performance analysis framework of the autonomous distributed computing system such as Grid computing and web services using OPNET modeling and simulation. It is necessary to enable wide-spread sharing coordinated use of network and server resources for IT systems to become responsive, variable, and autonomous. It is needed to integrate distribute, heterogeneous, virtual organization in order to provide convenient services. We consider the generic service-oriented architecture that consists of clients, resource manager, job/task scheduling manager, and application managers. In this architecture, there are two key protocols. One is the lookup protocol. It discovers the appropriate application servers. The other is the dynamic feedback protocol. It monitors server congestion state remotely and sends the report to the resource manager. We modeled these protocols on OPNET. The developed simulation modules for these protocols enable us to evaluate the performance of the services based on the service-oriented architecture. The case studies using these simulation modules those are applied to the superscheduling service and the adaptive application agent service are shown in the next our presentation.