Distributed operating system for resource discovery and allocation in federated clusters

Nowadays, numerical simulation has an important place in several scientific fields where real experimentation is expensive or impossible. Important computing resources must be used to realize these experimentations and workstations are not suitable to satisfy these requirements. After having aggregated computers in clusters or in federated clusters (or grid), a challenge for laboratories or large companies is to efficiently and easily use this large amount of resources. Several projects like PBS [Bode00] or Kerrighed [Morin04] offer solutions for cluster resources management. Other projects like Condor [Litzkow88], Globus [Foster97] or NetSolve [Agrawal03] focus on federated clusters or grid computing.