Traditional Cloud model is not designed to handle latency-sensitive Internet of Things applications. The new trend consists on moving data to be processed close to where it was generated. To this end, Fog Computing paradigm suggests using the compute and storage power of network elements. In such environments, intelligent and scalable orchestration of thousands of heterogeneous devices in complex environments is critical for IoT Service providers. In this vision paper, we present a framework, called Foggy, that facilitates dynamic resource provisioning and automated application deployment in Fog Computing architectures. We analyze several applications and identify their requirements that need to be taken intoconsideration in our design of the Foggy framework. We implemented a proof of concept of a simple IoT application continuous deployment using Raspberry Pi boards.