Smart farming is an innovation in the agricultural sector that can improve the performance of the agricultural industry with more automated and datadriven farming methods that connect with the IoT devices in the farming area. The fast spread of connectivity has given rise to IoT-based agricultural management solutions. Most of the existing farming systems, which are designed using the traditional cloud computing architecture, are unable to handle massive volumes of data produced by the connected IoT devices and may have high latency upon heavy traffic. Consequently, it is preferable to bring the data processing closer to the source of its production in order to minimize the latency and network usage in assisting real-time decisions based on the data generated. This shows the deployment of application modules is important for the efficient utilization of network resources. For this reason, fog computing model has been proposed in this paper to solve such deployment issues. The architecture is to be designed for smart farming, and simulated using iFogSim, which enable to evaluate the usage of bandwidth and computing resources as well as latency. The performance results obtained by the proposed fog computing approach are to be compared to those of cloud-only implementations in order to give recommendations for practical use.