Dynamic Preclusion of Encroachment in Hadoop Distributed File System

Big Data has been one of the current and future research frontiers. Data driven decisions derived from big data have critical significance in many application domains, e.g., health, commerce, finance, marketing, military, etc. In big data, computations have to operate on massive data sets to extract actionable intelligence and storage system must also support concurrent computations. The Hadoop file system has been designed to meet the complex requirements of big data. High value associated with big data sets has also rendered big data storage system attractive target for cyber attackers. In this work, we focus on a process specific security model for HDFS in big data to secure the data on various operations in nodes of HDFS. The proposed model effectively precludes the intruder prone operations such as data deletion, insertion and replacement in the cluster nodes which pose severe threat to the HDFS functionality using Apache Sentry for authorization of Clients and Kerberos for authenticating access to the nodes.