Stochastic models of multiprocessor systems: stability and moment properties

The basic models of multiprocessor systems in which each job requires a random number of processors have been studied. A new model of such system is proposed and studied, where, for a given job, all processing times are identical. This assumption reflecting real service process makes the analysis of the system more difficult. To investigate the model, some minorant and majorant (classical) systems are constructed, which allow, for a few important particular cases, to obtain both stability conditions and the moment properties of the original stationary workload process.