Ever increasing energy consumption has been a major cost for maintaining data centers, and there is urgent need to reduce the energy consumption of data centers. The key components contribute a large portion of energy consumed in a computing node, both in terms of electrical cost and cooling cost. In this paper, we studied the performance and thermal profiling of a cluster, and proposed an approach to model energy consumption of a computing node based on the workload that is dispatched to the node. Statistical methods were used in generating the temperature and computing energy cost model, and our experimental result indicated high accuracy of our temperature model and energy model. Our proposed approach could be extended to large-scale cluster systems and data centers.