Nowadays, the use of IoT devices is widely standardised in different areas of our lives, from work, home, educational institutions, etc. Such amount of sensing data needs to be controlled in an efficient way, and FOG computing is one of the most used paradigms in this kind of environments, due to its ability to solve the computational needs of data generating devices, reducing the traffic between them and decentralising the data processing. Due to the constant growth of IoT environments, the FOG layer must be able to configure and scale as the infrastructure grows, adapting to the new computational requirements of the environment. These operations are still performed manually by system administrators, resulting in an inefficient and error-prone practice. This work proposes the design and development of a FOG layer capable of dynamically controlling the configuration of the logical components of the layer, providing the infrastructure with new nodes, a coordinator and a balancer, alongside the existing FOG processor nodes. The coordinator node balances the load distribution for the processor nodes, while the balancer will perform the load distribution under the coordinator’s guidelines. This proposal automates the operation of IoT infrastructures, provisioning with the minimum resources to cover the computational needs. To validate the proposal, the model has been deployed on the campus of the University of Alicante as a FOG layer of the sensorisation system.