Developing software systems to Big Data platform based on MapReduce model: An approach based on Model Driven Engineering

Abstract

Context: The need to analyze a large volume and variety of data for the purpose of extracting information has been promoting investments in Big Data, e.g., for storage, analysis and, more recently, methodologies and approaches for software system development for Big Data platforms. The application of software engineering for Big Data is recent and emerging, so in the literature we find a number of challenges and opportunities related to Big Data, but few practical approaches. ObjectiveIn this paper, we propose a practical approach based on MDE (Model Driven Engineering) to support the semi-automated development of software systems for Big Data platform that use MapReduce model. MethodThe proposed approach consists of framework, process, metamodels, visual Alf, transformation definitions written in ATL and Eclipse IDE plug-in. The proposed framework uses concepts of MDE, Weaving and software development based on Y. Our proposed process guides the use of our approach. A graphical notation and extended metamodel for Alf (i.e. visual Alf) assign executable behavior for UML or DSLs. An Eclipse IDE plug-in implements our approach. ResultsWe show the applicability of the proposed approach through an illustrative example. ConclusionOur approach brings a contribution because the development of software systems is assisted by models which preserves the business logic and adds Big Data features throughout the development process.