Optimal population mix in pool depletion systems with two-class workload

The evolutions of digital technologies and software applications have introduced a new computational paradigm that involves the concurrent processing of jobs taken from a large pool in systems with limited capacity. The definition of admission control policies that choose which jobs to process is crucial to improve the overall performance especially in systems with multiclass workload. In a previous work we show that in such systems, hereinafter called pool depletion systems, few parameters have a non-trivial impact on the processing time of the whole pool. Other performance indices, such as the energy consumption, are also deeply affected. In the present work, we further investigate such phenomenon by applying results from queueing theory, absorption time analysis and by performing discrete event simulations. We propose different techniques in order to identify the optimal or near-optimal setting. We analyze their complexity and provide guidelines to choose which of them adopt according to the application scenario characteristics.