Although the main objectives of dynamic spectrum access and cognitive radios are to solve the problem of spectrum scarcity and usage inefficiency, we show in this paper that such capabilities can also be used to improve other performance metrics in wireless networks. We compare the performance of a Cognitive Radio (CR) network using nodes that are able to sense the spectrum and change their operating frequency in case of link failure with a Traditional Radio (TR) with static channel access and no frequency switching capability. The improvement that can be obtained by substituting TR to CR technology is evaluated for the blocking probability, utilization and system occupancy performance metrics based on a Two-state Markov chain channel model as a function of the number of users, number of channels, arrival rate, restoration time and channel occupancy. We show that there is a strong interaction between these parameters and that CR technology performs better in slowly time-varying channels while TR technology is more appropriate in fast changing channels.