The authors present a data block replication method to increase energy efficiency in data centers. The paper starts with a clear introduction and background section, which provides simple explanations about the basic concepts. This section describes the target Hadoop distributed file system, the baseline power down strategy (covering subset), and details of the server-sharing delay scheduling algorithm. Even a very naive reader can understand the remaining content after reading these preliminaries. The proposed work and algorithms are also very straightforward and easy to follow.

While experimental results reveal that the proposed policy performs better for the optimal resource utilization point at which the energy consumption of the servers is minimized, the paper does not provide any concrete energy measurement. I expect to find a detailed energy consumption analysis in a work that tries to optimize energy efficiency. The tradeoff between performance and energy would have been more evident if the authors had presented more precise energy results.