Cloud-Based Distributed Mutation Analysis

Abstract

Mutation Testing is a fault-based software testing technique which is too computationally expensive for industrial use. Cloud-based distributed computing clusters, taking advantage of the MapReduce programming paradigm, represent a method by which the long running time can be reduced. In this paper, we describe an architecture, and a prototype implementation, of such a cloud-based distributed mutation testing system. To evaluate the system, we compared the performance of the prototype, with various cluster sizes, to an existing "state-of-the-art" non-distributed tool, PiT. We also analysed different approaches to work distribution, to determine how to most efficiently divide the mutation analysis task. Our tool outperformed PiT, and analysis of the results showed opportunities for substantial further performance improvement.