Implementation and performance evaluation of the parallel CORBA application on computational grids

Abstract

The grid computing technology provides a new solution to high performance distributed computing. Computational grids can provide dependable, consistent, pervasive and inexpensive access to high-end computational capabilities to meet requirements of high performance scientific and engineering applications. Algorithme génétique 2-dimensionnel (Ag2D) is an application for airfoil shape optimization. The application is configured in the client-server architecture. Parallel genetic algorithm is employed in the optimization process. In the implementation, a collaborative computing environment, CAST, is used to define modules and workflow of Ag2D application. Multiple modules of Ag2D are implemented in parallel CORBA (PaCO) objects. Inside the PaCO object, Ag2D application is implemented with message passage model. CORBA run-time system works as the executing environment for Ag2D application. Unicore, a grid middleware, functions to deploy various PaCO services, and manage resources/tasks. Unicore thus takes the role of supporting environment of Ag2D application. Test results and performance evaluation justify the contributions – efficiently implement the parallel object application on computational grids and exploit traditional parallel/distributed programming models in grid environments.