Policy Driven Node Selection in MapReduce.

Anna Cinzia Squicciarini,Dan Lin,Smitha Sundareswaran,Jingwei Li

DOI: https://doi.org/10.1007/978-3-319-23829-6_5

2014-01-01

Abstract:The MapReduce framework has been widely adopted for processing Big Data in the cloud. While efficient, MapReduce offers very complicated (if any) means for users to request nodes that satisfy certain security and privacy requirements to process their data. In this paper, we propose a novel approach to seamlessly integrate node selection control to the MapReduce framework for increasing data security. We define a succinct yet expressive policy language for MapReduce environments, according to which users can specify their security and privacy concerns over their data. Then, we propose corresponding data preprocessing techniques and node verification protocols to achieve strong policy enforcement. Our experimental study demonstrates that, compared to the traditional MapReduce framework, our policy control mechanism allows to achieve data privacy without introducing significant overhead.