Achieving ultra rapid evolution using service-based software

There is an urgent industrial need for new approaches to software evolution that will lead to far faster implementation of software changes. Existing software maintenance processes are simply too slow to meet the needs of many businesses. To achieve the levels of functionality, flexibility and time to market of changes and updates required by users, a radical shift is required in the development of software, with a more demand-centric view leading to software which will be delivered as a service, within the framework of an open marketplace. Although there are some signs that this approach is being adopted by industry, it is in a very limited and restricted form. We see ultra rapid evolution, in "internet time" as a grand challenge for software engineering.In this position paper, we describe recent work that has resulted in an innovative demand-led model for the future of software. We describe a service architecture in which services may be bound instantly, just at the time they are needed and then the binding may be disengaged. Such ultra late binding requires that many non-functional attributes of the software are capable of automatic negotiation and resolution. Some of these attributes have been demonstrated through two prototype implementations based on existing and available technology. The aim of the position paper is to contribute to the debate at ESEC by presenting a radical, market based view of software evolution which must take place in "internet time". The key underpinning theoretical idea is ultra-late binding, so that a service is engaged dynamically at the point in time it is needed. Hence, the sub services can evolve between usages.

[1]  Keith H. Bennett,et al.  A Staged Model for the Software Life Cycle , 2000, Computer.

[2]  David Griffiths,et al.  The future of software , 1999, CACM.

[3]  Pearl Brereton,et al.  Service-based software: the future for flexible software , 2000, Proceedings Seventh Asia-Pacific Software Engeering Conference. APSEC 2000.