Hadoop-based Storage Strategy for Mass Files
Jiang Jianj · Journal of Shenzhen Polytechnic · 2014
In order to solve the performance bottleneck of small files I/O in classic HDFS, this paper proposes an optimal HDFS storage strategy characterized by file preprocessing, metadata information expansion, and improved metadata mass operation. Tests indicate that the optimal strategy can improve the mass small files I/O performance for HDFS.