Over the past few years, network coding has been employed in data reconstruction process of storage systems to minimize the recovery bandwidth. However, the time consumption of the decoding operations incurs a significant performance degradation. In this paper, we propose a data temperature-based reconstruction optimization algorithm and integrate it into the reconstruction process of a Network-Coding-Based File System (NCFS) which adopts regenerating code as its storage coding scheme. We conduct extensive experiments to evaluate the impacts on the data reconstruction process of regenerating codes. The experimental results demonstrate that our method outperforms the conventional approach both in reconstruction time, throughput and average response time with up to 33.17%, 60.61%, 37.77% improvement, respectively.