Determination of parallel software organizations for performance-constrained and connectivity-constrained distributed systems

The process connectivity constraints and the resulting impact on the computation structure are considered. Within these constraints, an approach is presented for restructuring a set of subtasks with precedence constraints and stochastic properties into alternative parallel structures that can be scheduled in parallel distributed systems. A queuing model is used to evaluate the module response time and its resource cost. This model takes into consideration both the communication and the synchronization delays. Finally, an algorithm for determining feasible schedules (that meet designer performance constraints) in order of minimum resource cost is developed.<
>