Sustainability-Aware Cloud Computing Using Virtual Carbon Tax

In this paper, a solution for sustainable cloud system is proposed and then implemented on a real testbed. The solution composes of optimization of a profit model and introduction of virtual carbon tax to limit environmental footprint of the cloud. The proposed multi-criteria optimizer of the cloud system suggests new optimum CPU frequencies for CPU-cores when the local grid energy mix or the cloud workload changes. The cloud system is implemented on a blade system, and proper middlewares are developed to interact with the blades. The experimental results show that it is possible to significantly decrease the targeted environmental footprint of the system and keep it profitable.