Nowadays, the energy production from solar radiation becomes more important in the light of current environmental challenges. Current research aims at combining solar energy generation with urban planning in order to maximize efficiency. The goal is to evaluate the potential of building roofs located in urban areas for producing solar energy. This paper deals with a decision support system that calculates the solar energy potential of surfaces in urban landscapes. The system should identify "good candidate" roofs for installing solar panels. It should quantify the solar power a roof could generate by taking into account local weather, roof's orientation and how much shade falls on it from nearby trees and buildings. The system is based on a CPU and memory intensive "shadow process" algorithm. We propose two paradigms to describe this algorithm. The first is based on a conventional parallelization. The second paradigm is based on an optimized data distribution. The two paradigms have been implemented on cloud computing infrastructures. The paper compares the two methods on the basis of two criteria: performance and deployment cost.