<span lang="en">Packet capturing and analysis in high-speed networks, like 1-10 Gigabit Ethernet, is a challenging task, especially when applications do not permit packet loss. In this environment, the use of multiprocessor and multicore systems, as well as the parallelization of applications, is aimed at improving the performance. However, the monitoring application may even experience performance penalties when adapted to multiprocessor architectures. After observing certain anomalies in a real traffic monitoring system that runs on a multiprocessor platform, this paper presents an analytical model for that type of systems. The model, which is based on generalized stochastic Petri nets, evaluates the efficiency of the traffic capturing and analysis system depending on the hardware/software platform features.</span>