A scheduling advisor for distributed systems

An adaptive scheduling advisor is proposed for computing resource management in distributed systems by providing a means for optimal mapping of an application onto the system. To make its decision, the scheduling advisor uses the job information, a detailed description of the system, and current dynamic parameters. The decision-making process is controlled by a set of scheduling heuristics, a simple rule base, and a set of cost functions. In addition to fixed-size jobs, it is able to produce schedules for jobs that can be scalable over a variable number of nodes.<
>