Model-Driven Software Development

Software development is a complex and difficult task that requires the investment of significant resources and carries major risk of failure. According to its proponents, model-driven (MD) software development approaches are improving the way we build software. Model-driven approaches putatively increase developer productivity, decrease the cost (in time and money) of software construction, improve software reusability, and make software more maintainable. Likewise, model-driven techniques promise to aid in the early detection of defects such as design flaws, omissions, and misunderstandings between clients and developers. The promises of MD are rather lofty, and so it is only natural to find many skeptics. As Brooks famously described [Bro95], software engineering will not likely deliver the sort of productivity gains we experience in hardware engineering where we see “Moore’s law”-styled doublings every 24 months [Moo10]. Thus, if we accept Brooks’ premise, nobody should expect any innovative approach to software development to be a “magical silver bullet” that will increase productivity by an order of magnitude within a decade. Unfortunately, the amount of hyperbole surrounding the various flavors of MD sometimes makes it seem like advocates believe MD to be a silver bullet. Model-driven development is no panacea. However, we believe that model-driven is a superior approach to software construction. This chapter examines the current state of the art in model-driven software development. We begin by characterizing the various approaches to model-driven development (Section 2). Then we examine what modeling is and why we engage in modeling (Section 3). Next, we explore the history of software modeling that has led to current model-driven approaches and discuss what is required