Due to the increasing proliferation and serious resource constraints of mobile devices, using the rich-resource server can help the mobile client to perform some tasks for improving performance and saving resources is more and more needed in pervasive computing. Remote execution is a static software configuration approach that is unsuitable for constantly changing environments. In this paper, we present a dynamic software partitioning algorithm that can allocate optimally the appropriate components among the machines to minimize the specified resource consumptions. Since dynamic software partitioning often leads to components? migration, we consider component migration cost, plus computation cost and communication cost, to set up a generic math model of resource consumption. Based on the network flow theory, we reduce the optimal software partitioning problem to the min-cut problem of a cost graph. Our algorithm can be used to find an optimal software partition to conserve network bandwidth, decrease response time of services and save energy of mobile devices, respectively. The simulation results show the algorithm is very effective, and can save significantly more resource than the previous studies.