MapReduce: A critical analysis of existing authentication methods

Abstract

The use of MapReduce (MR) model has attracted much attention because of the reductions in the costs involved in satisfying the storage and computational demands. However, authentication methods proposed for such model is subjected with rooms for improvements. In the initial design and implementation of the MR model such as Hadoop, issues of security including authentication was not considered. It was assumed that the underlying system worked in a trustworthy environment, so MR developers did not provide authentication methods in the MR model. This assumption is still valid to some extent, whereas major IT players in the world, such as Facebook and Yahoo have implemented the MR model in their private clouds. However, recently there are efforts to provide the MR model in public clouds. Inadequate authentication methods for the MR model in such environment makes the job execution work-flow vulnerable to many attacks such as impersonation and spoofing attacks. So as part of our work to design a secure and effective authentication solution for the MR model, this paper examines and critically analyses the state-of-the-art in MR authentication methods against a general set of specified requirements. It is to identify limitations, weaknesses, and missing features in the proposed authentication methods for the MR model.