Came across this nice paper - also can be seen as a primer on SaaS. I do not generally get the time and patience to study and recommend academic papers - this one somehow stood up for its simplicity - though some ideas presented here are academic, looked like a good outline. Excerpts with edits:
- Software will need to be developed to meet necessary and sufficient requirements,i.e. for the majority of users whilst there will be a minimum set of requirements software must meet, overengineered systems with redundant functionality are not required. For example, users of a sophisticated word processor may only need a very small subset of its capabilities and, from the user’s point of view, should only need to acquire and pay for that subset.
- Software will be personalised. Software is currently packaged and marketed as a generic product with little scope for configuration or personalisation. In future, software will be capable of personalisation, providing users with their own tailored, unique working environment which is best suited to their personal needs and working styles, thus meeting the goal of software which will meet necessary and sufficient requirements.
- Software will be self-adapting. Software will contain reflective processes which monitor and understand how it is being used and will identify and implement ways in which it can change in order to better meet user requirements, interface styles and patterns of working. It will also identify the need to commission new or changed software and decommission redundant software as and when user requirements change and thus supporting personalisation.
- Software will be fine-grained. Future software will be structured in small simple units which co-operate through rich communication structures and information gathering. This will provide a high degree of resilience against failure in part of the software network and allow software to re-negotiate use of alternatives in order to facilitate selfadaptation and personalisation.
- Software will operate in a transparent manner. Software may continue to be seen as a single abstract object even when distributed across different platforms and geographical locations. This is an essential property if software is to be able to reconfigure itself and substitute one component or
network of components for another without user or professional intervention.
- The rental model is based upon the rent or hire of software from a producer, as a means of reducing upfront costs. For example, more than half of UK companies are planning, within the next year, to use services that allow them to rent certain software items rather than buying them . Strictly, the rental model does not imply any change to the physical structure or installation location of software, and so is merely a change in payment method.
- The server model is based upon the use of thin clients to offer software from a central server with a charging regime based on pay-per-use, typically to avoid upfront procurement costs by user organisations and achieve upto-the-minute maintenance through access to the latest release of software. However this model does not necessarily require any change to the basic structure of the software and relies on achieving user flexibility through the distribution network. The problem of maintenance and delivering flexibility
- The service package model is based on a well established trend for products to be packaged with a range of services designed to support and enhance product use. For example, an airline offering seats as its core product, may offer a range of additional, value adding services as a package. Similarly, some software producers offer business solutions comprising product and service elements. Again, this concept does not imply any change in the nature of the underlying software product itself, although users may be provided with different experiences through the service layer surrounding the product.
The ability to change software to meet new business needs (i.e. software evolution) improves by only a few percentage points each year. Even very radical projections of current technologies, tools and processes (for example, COTS, reuse etc) seems unlikely to accelerate this rate very significantly. Yet businesses developing in internet time need orders of magnitude improvement in change rates to bring ideas to market far faster than is currently possible.