This paper presents an evaluation of Finite Control Set-Model Predictive Control (FCS-MPC) performance for dynamic resources provisioning. Realization of a shared-resources environment is experimented in software system network by virtualization technique. Control parameter regulates their sources provisioning among the users to guarantee the achievement of performance metrics in the network. The objective preferences are maintained in relative term by manipulating a finite set of operating points as the control states. Moreover, an optimization procedure is implemented based on receding horizon prediction of MPC in a feedback control scheme. Further investigation is conducted to examine the performance of optimal FCS and constrained FCS for dynamic system management. The experimental results demonstrate that constrained FCS system provides a better response stability than the optimal FCS. Correspondingly, the range and quantity of input states are significantly influential to the flow of control action. A wider range and longer control state sequence yields to a smoother control signal stream.