MATRIX-HPC Preliminary Results Architecture Execution Unit

• Efficiently scheduling large number of jobs over large scale distributed systems is very critical. • Today's state-of-the-art job schedulers mostly follow a centralized architecture that is master/slave architecture. • Aims at providing HPC support on top of MATRIX MTC framework. • In-corporates resource stealing with work stealing. • Scheduler chooses multiple nodes in random. • Requests for resource information on the nodes. • Validates if sufficient resources are available to complete the tasks. • If Yes, Breaks the task into sub-tasks and migrates it to the nodes selected. • Source receive the results after execution Abstract Workin