Brunel University's Business Class: Building a model information system

Model driven architecture can sound complicated but this month Professor Robert Macredie and Dr Mark Lycett explain how this evolution is really about making IT easier - and more effective...

The IT industry has, for some time, been talking about moving towards a model driven approach to software development. Now, with the upsurge in outsourcing development, programming is becoming a commodity and the debate is resurfacing.

Outsourcing software development demands a very specific understanding of requirements. This is where a model driven approach comes in as it elevates the task of understanding business systems and modelling them. The question is whether moving towards this approach to software development is inevitable? And if so, what are the ramifications of the change?

For those less focused on the more theoretical aspects of IT, model driven architecture (MDA) is an example approach to model-oriented software development espoused by the Object Management Group. The approach effectively separates the business logic behind a system from the technology used. An organisational ‘blueprint’ of business systems is created, which can then be translated to specific software models.

The theory is that this method of development can be used to create platform-independent applications, which are so abstracted from the code that they are insulated from the effects of technological change. Any business change can be discussed with business people, then later translated into a technological solution (offering the potential of doing this with greater speed and less effort).

Equally, if technological change is necessary, the blueprint will still remain the same, so there would be no need to redesign the system from the organisational viewpoint, purely from the technological side.

Delving further into the theory, the Object Management Group’s idea of model driven architecture comprises three interlinked viewpoints:

Computational independent, which is intended to show the business semantics, namely what it is the business does. Any detail of IT structure and processing is hidden at this stage.

Platform independent, which focuses on the operational aspects, while hiding the details necessary to ally the work to a particular development platform.

Platform specific, which supplements the platform independent viewpoint with a focus on the specific technology platform: the last piece of the jigsaw.

The benefits of adopting this model driven approach are immediate: the early stage of development (the computational independent part) is understandable to business people. This means that the initial development and communication of project requirements is a lot easier for both business and IT departments. It may even help to bring an end to the business/IT divide – at least in the area of project development.

However, the disadvantage is that, in order to successfully move to this approach, there is a lot of work to do. It raises the question of whether we are set up to manage this process effectively.

It is going to require a considerable investment of time for a business to reach an understanding of what exactly it does and how IT can help it do better. Ask any person exactly what it is they do and the answer is, at best, a long one. At worst, it’s unintelligible or punctuated by long silences.

Trying to communicate that picture in a wider business sense will be a real challenge – but it’s where 80 per cent of the work involved in business development lies and it’s absolutely crucial to the success of a project. Understanding the semantics of business may not sound like an important task but examine closely the cost and effort necessary to rationalise different IT systems conceptualisations of ‘customer’, for example. (Generally not a trivial task.) Typically, business understanding has concentrated on ‘how’ and not ‘what’, which is where the least research has been done to date.

The ambitious theory behind this approach to software development is that, once the approach is mature, each stage of the process could be automated to some degree. If this is realised, then business changes could be implemented by modifying the business model, after which the system could be automatically regenerated.

This truly is a picture of a brave new world of software development. However, the question is not whether we are brave enough to try to turn it into reality – research teams are now turning their hand to working on this. The real challenge lies in whether we can break down the important aspects of business sufficiently to understand what it is we do and translate that into what we need from an information system.