H-RCBAC: Hadoop Access Control Based on Roles and Content

Abstract

Social networks, smartphones, mobile applications... produce an avalanche of data on a large-scale and in an unstructured way. The phenomenon Big Data was born in order to address the different challenges including data storing, data analysis, data querying and so on. Technological advances always carry new security vulnerabilities that are not taken into consideration at the beginning. Security aspects usually require time to be addressed. The information system security is the set of measures to prevent any failure or threat including unauthorized accesses. Perfect protection must contain the four basic building blocks that are: authentication, access control, auditing, and encryption. In our work, we are specifically interested in access control. We first analyze the well-known access control models that were applied to Big Data. We then investigate the most important security projects. Most of these approaches and projects rely mainly on coarse-grained access control policies. In this work, we propose a novel approach called H-RCBAC that relies on two known models: the role-based access control (RBAC) and the content-based access control (CBAC). H-RCBAC is a new architecture that refines the access control process by considering a set of taboo words to guarantee fine-grained access control.