The efficient utilization of resources is an issue of essential importance in modern network management. The emergence of increasingly complicated and continuously changing network services requires softer and fuzzier methods in network management. In this paper we demonstrate how evolutionary game theory can be used for an efficient allocation of service requirements onto an ensemble of heterogeneous network components. By incorporating differentiated pricing structures into a system utility function, network agents are encouraged to increase their usage of those components that are presently badly utilized. It is demonstrated how this approach can enhance network utilization significantly. Some new results regarding evolutionarily stable strategies in nonlinear evolutionary games are also reported.