Composite Scheduling Strategies in Distributed Computing with Non-dedicated Resources

This work presents dispatching strategies based on methods of job-flow and application-level scheduling in virtual organizations of distributed computational environments with non-dedicated resources. Job-flow management is implemented with the set of specific rules for resource usage. Applications are considered as parallel jobs. Strategies are based on economic scheduling models and diverse administration policies inside resource domains (clusters, computational nodes equipped with multicore processors etc.). Methods of priority economic scheduling of global job flows and local-level applications in distributed computations are studied. Job management structures and economic mechanisms for load balancing in distributed environments are considered.