Fog computing extends cloud computing technology to the edge of the infrastructure to support dynamic computation for IoT applications. Reduced latency and location awareness in objects’ data access is attained by displacing workloads from the central cloud to edge devices. Doing so, it reduces raw data transfers from target objects to the central cloud, thus overcoming communication bottlenecks. This is a key step towards the pervasive uptake of next generation IoT-based services. In this work we study efficient orchestration of applications in fog computing, where a fog application is the cascade of a cloud module and a fog module. The problem results into a mixed integer non linear optimisation. It involves multiple constraints due to computation and communication demands of fog applications, available infrastructure resources and it accounts also the location of target IoT objects. We show that it is possible to reduce the complexity of the original problem with a related placement formulation, which is further solved using a greedy algorithm. This algorithm is the core placement logic of FogAtlas, a fog computing platform based on existing virtualization technologies. Extensive numerical results validate the model and the scalability of the proposed algorithm, showing performance close to the optimal solution with respect to the number of served applications.