Collaborative DFA learning applied to Grid administration

This paper proposes a distributed learning mechanism that learns patterns from distributed datasets. The complex and dynamic settings of grid environments requires supporting systems to be of a more sophisticated level. Contemporary tools lack the ability to relate and infer events. We developed an information system, based on collaborative agents, that supports system administrators in monitoring the grid. While observing log files, the agents learn traffic patterns in their own local domain of the grid. The agents represent their knowledge in the form of deterministic finite automata (DFA), and share their models to provide global or multi-domain overviews. We discuss our collaborative learning mechanism and show the results of our experiments with data of two grid-sites. Our system generated job-traffic overviews that gave new insights in the performance of the grid environment.