Víctor Méndez Muñoz (Universidad de Zaragoza)
Felix García Carballeira (Universidad Carlos III de Madrid) Data grid replication is critical for improving the performance of data intensive applications. Most of the used techniques for data replication use Replica Location Services (RLS) to resolve the logical name of files to its physical locations. An example of such service is Giggle, which can be found in the OGSA/Globus architecture. Classical algorithms also need some catalog and optimization services. For example, the EGEE DataGrid project, based in Globus open source components, implements for this purpose the Replica Optimization Service (ROS) and the Replica Metadata Catalog (RMC). In this paper we propose a new approach for improving the performance of Data grid replication. With this aim, we apply Emergent Artificial Intelligence (EAI) techniques to data replication. The paper describes a new algorithm for replica selection in grid environments based on a PSO-LRU (Particle Swarm Optimization) approach. For evaluating this technique we have implemented a grid simulator called SimulaGrid. The simulation results presented in the paper demonstrate that the new technique improve the performance compared with traditional solutions. Data Grid infrastructures , Data-intensive Grid applications , Data Grids, Web services and P2P , Indexing, caching and load balancing in Grids , Grid performance evaluation

Download the fullpaper