Enhanced multilevel security for hadoop distributed file system using image steganography

Abstract

In the Internet era, there is an exponential increase in the amount of data newlineproduced from various sources, including social networks, sensors, the Internet newlineof things, retail, logistics, financial databases, etc... The data is produced in a newlinevariety of structures like text, image, audio, video, etc. at high speed and this data newlineis called big data. The traditional RDBMS software can t handle such a large newlinevolume of various structures of data. Hence, separate storage models and newlineprocessing tools are needed for handling big data. Hadoop is a de facto standard newlinetool being widely used for storing and analyzing large volumes of data. Hadoop newlineis an open-source distributed framework that provides distributed storage and newlineparallel processing using commodity hardware. Hadoop has a storage platform newlinenamed Hadoop Distributed File System (HDFS) for storing big data and newlineMapReduce for processing. When big data was introduced, the primary focus newlinewas given to storage and processing of large volumes of data, the security of the newlinestored data was not given precise concentration. Since Big Data consists of newlinehighly sensitive data like electronic health records, financial data, etc., harnessing newlinebig data access and security plays a vital role. newlineIn four aspects an information system can be secured. They are newlineauthentication, authorization, encryption and auditing. In Hadoop clusters, the newlineusers are not authenticated either at NameNode or DataNodes where data is newlinestored. The authorization is done by the NameNode. Last layer of defense can be newlineprovided by encryption. There is no default encryption algorithm to secure the newlineHadoop Distributed File System. A periodic auditing of log file will help to find newlineintruders. But Hadoop log files consisting of HDFS transaction logs are not newlineaudited. newline