摘要
New experiments including those at the LHC will require analysis of very large datasets which are best handled with distributed computation.We present the design and development of a prototype framework using Java and Objectivity.Our framework solves such analysis-specific problems as selecting event samples from large distributed databases.producing varialbe distributions,and negotiating between multiple analysis service providers.Examples from the successful application of the prototype to the analysis of data from the L3 experiment will also be presented.
