Secure HDFS Design Using Group Key Management System

Abstract

Data is increasing in recent years due to the development of mobile communication technology and the spread of smart devices. These large amounts of data are called ＂Big-Data.＂ To process Big-Data, a variety of technologies have emerged. Of those, the most prominent technology is Hadoop. Hadoop introduced Kerberos and Token system for security. Nonetheless, it still has various problems including session key leakage issue, key sharing issue of NameNode and DataNode, replay attack of block accessed token, impersonation attack, etc. This study proposed HDFS using group key management system in order to resolve these vulnerabilities while keeping the optimal performance. The proposed system manages the secret key and group key of client and the group key of DataNode by bundling clients and DataNode as a respective logical group. Also, our study proposed client authentication using the proposed system and authority verification and BAT protocol when accessing files.