This paper investigates the class of Generalized Stochastic Petri Nets (GSPN) as a convenient and powerful analytical framework for the integrated modeling of the logical and time-based dynamics of sequential resource allocation systems (RAS). It is shown that the standard GSPN theory provides an exact formulation for the problem of performance control of logically controlled RAS with exponentially distributed event firing times, and effective approximations for RAS with more general timing distributions. This formulation can be subsequently utilized to (re-)establish various properties of the optimal scheduling policy for the considered class of systems, including the existence of a deterministic stationary optimal scheduling policy. Finally, the presented framework is explicitly solvable for small-sized RAS, and therefore, it can offer a more systematic study and a more profound understanding of the timed dynamics taking place in these environments.