Grid computing is a form of computing on heterogeneous, distributed systems more and more used to run scientific applications. These large scale experiments require an easy, automatic and efficient way of selecting resources, and then installing and running the applications on them, in order to allow users not experts in computer sciences to access this computing power. Today's tools do not cover the whole deployment process, the biggest lack lies in planning algorithms. Therefore, it becomes mandatory to provide new efficient tools. Among the deployment softwares, Adage seems the most promising one. We aim at implementing new deployment algorithms in Adage. The problem we are dealing with is to find a deployment which minimizes the communications between the applications, balances the load among the processors, and maximises the number of applications' instances on the platform.
You can download my master thesis here: in French or in English.