Several solutions have been proposed for Mobility Management in the Internet, ranging from the Network to the Application Layer. Existing support for mobility management is designed to work either at the lower protocol layers (L2 and L3), at the upper layers (L4, L5), or even across layers. In this paper we consider handling the mobility from the perspectives of the upper protocol layers and evaluate an implementation of this solution in comparison with relevant solutions implemented at the lower layers. This paper shares our experience in prototyping applications which tolerate connection disruptions caused by user's mobility. We also discuss, in terms of implementation, how the well-known mobility techniques should interact to handle the effects of handovers at the application level. Results obtained from experiments in a testbed are presented and compared with the performance of the Mobile IP.