Arrival and service curves are core functions in the Network Calculus framework. Based on those curves, we present in this talk a new formalism for data packetization in Network Calculus, the packet curves. Indeed, a more precise knowledge of the packet characteristics can be efficiently exploited to get tighter performance bounds, for example for aggregation of flows, packet-based service policies and shared buffers. Finally, we will give a model for a wormhole switch and show how our results can be used to get efficient delay bounds.