The increasing number of competitors and the growing traffic demand are the main factors pushing for a more dynamic and flexible service demand allocation mechanism. Human interactions are becoming more and more inadequate to solve this task, since many different issues have to be considered for optimising all possible choices and strategies. Software tools are becoming fundamental for supporting human decisions and/or for reducing the need of human intervention. Nowadays, agent technology promises a good support for pro-active and autonomous network control, that would enable the automation of many network provider tasks. In order to prove the feasibility of such automation through the use of agents, a multi-agent simulator for the allocation of service demands has been developed. This paper describes the simulator and aims to give a useful feedback for agent developers.