Abstract-Model-Driven Software Engineering (MDSE) has put a great emphasis on modeling to deal with the rapidly growing complexity of software systems. This leads to that models become more complex and detailed as they evolve. In contrast to the vision that such models are able to deal with this complexity, there are several cases in which simple models better support the requirements. In this paper, three technology trends are presented. The success of these trends lays in using lightweight models. Of course, inherently complex systems tend to use complex models but the main goal is to focus modeling on supporting requirements and avoid adding unnecessary details. After the explanation of the technology trends, their key success factors are identified. By these success factors, we outline some considerations that facilitate maintaining the complexity of models low.Index Terms-Modeling, model-driven software engineering, agility.
I. INTRODUCTION
Model-Driven Software Engineering (MDSE)is a widely applied method in software engineering, which uses the model as a first-class artifact in the engineering process. The model is leveraged in software development instead of focusing on algorithmic concepts. The Model-Driven Architecture (MDA) methodology, standardized by the Object Management Group (OMG) even tries to avoid programming in the conventional sense and use modeling instead as a primary means of software development. The expansion of these technologies has emphasized the importance of the model. Nowadays, there is a big effort in creating more accurate and complete models. Often we ask ourselves, how we efficiently model problem domain D to capture all of the significant details.At the same time, complexity of software systems grows at a rapid pace. The result of this is that models become always bigger and more complex. Their purpose is to facilitate software development but big and complex models are very rigid and make future changes difficult. They also have a negative impact on scalability [3]. There is ongoing research on how to deal with distributed models and parallelized