In search of being able to offer better quality of service (QoS) in multiclass networks, a strategy that aims to allow resources between different classes of service (CoS) to be shared according to the class’s needs and priority is proposed. The goal is to achieve an optimal use of the total bandwidth in a link. This paper presents shortly the proposed strategy and explains the “squatting” and “kicking” mechanisms. A model and simple results for performance to prove their utility are shown.