Over the last decade, great efforts are invested in reducing emissions of greenhouse gases at a world scale. We should therefore elaborate new global solutions to this type of pollution, in which fleets of vehicles are heavily contributing. In this paper, we propose a novel green architecture for global fleet management that considers the most important actors. The proposed architecture allows to collect useful information from several fleets of vehicles and their environment, and provides real-time services to companies that own one of these fleets. One important service is the calculation of economic itinerary. We propose a new "green" calculation method that is based on an efficient and high precision fuel consumption estimation. Our method uses: (i) topological information such as the elevation variations, source and destination of the trip, (ii) weather conditions, (iii) vehicle's characteristics, and (iv) driver's behavior. We implemented a lightweight traffic management server that provides a real-time service allowing to find the most economic itinerary for vehicles belonging to the fleet. The utilization of such architecture and mechanisms associated with itinerary planning results in a highly green fleet management system.