Nowadays, energy saving has become a focus in deploying clouds. Developing energy-aware cloud data centers can not only reduce power electricity cost but also improve system reliability. Existing scheduling algorithms developed for energy-aware cloud data centers are commonly lack of consideration of task level scheduling. To address this issue, this paper proposes a novel rolling-horizon scheduling architecture for real-time task scheduling, together with a detailed task energy consumption model. Based on the novel scheduling architecture, this work develops an energy-aware scheduling algorithm EARH (energy-aware scheduling algorithm) for real-time aperiodic tasks. EARH employs a rolling-horizon optimization policy and can be extended to integrate other energy-aware scheduling algorithms. Strategies for the resource scaling up and scaling down are also presented to make a good trade-off between task’s schedulability and energy saving. Extensive experiments are conducted to validate the superiority of EARH by comparing it with three baselines. The results show that EARH significantly improves the scheduling quality over the others and it is suitable for real-time task scheduling in virtulized cloud data centers.