Multiple mass-market applications as components

Truly successful models for component-based software development continue to prove elusive. One of the few is the use of operating system, database and similar programs in many systems. We address three related problems in this paper. First, we lack needed models. Second, we do not know the conditions under which such models can succeed. In particular, it is unclear whether the notable success with operating systems can be replicated. Third, we do not know whether certain specific models can succeed. We are addressing these problems by evaluating a particular model that shares important characteristics with the successful operating system example: using compatible PC packages as components. Our approach to evaluating such a model is to engage in a case study that aims to build an industrially successful system representative of an important class of systems. We report on our use of the model to develop a computational tool for reliability engineering. We draw two conclusions. First, this kind of model has the potential to succeed. Second, even today, the model can produce significant returns, but it clearly carries considerable risks