The growing global demand for agricultural resources, combined with the need to reduce water and energy waste, is driving the adoption of smart systems based on IoT, cloud computing and decision-making automation. This is the context for the application analysed, which is designed to monitor agricultural fields using sensors, drones and actuators, with the aim of optimising irrigation and ensuring operational sustainability. However, its original implementation in an on-premise environment highlights structural limitations: unpredictable load growth, poor fault resilience, manual resource provisioning and infrastructure costs that are difficult to optimise. These critical issues highlight the need for a more flexible and automated architectural model. For this reason, the work explores the migration of the application to the cloud, focusing on two of the main cloud providers, namely Amazon Web Services (AWS) and Microsoft Azure. The dual modelling aims not only at comparative evaluation, but also at verifying the portability of the system, its vendor neutrality and the possibility of building a replicable, scalable and long-lasting architecture.