BCS / PGD/ MIS/ MIS Development (Approaches)

MIS Development

Traditional approach – development is undertaken in a series of steps, with each one being completed before the next starts. Steps may be requirements determination, design, coding, testing and installation. Maintenance can be seen as the next cycle, or the last of the series of steps. Another name used for the traditional approach is the waterfall approach, as it can be depict as a waterfall from one stage to the next in the development process. The motto of this approach is to “get things right first time” and backtracking to a previous stage is seen as weakness in the quality of the work within the previous stage. The approach is suitable for developments that have static and clear requirements at the outset of the development.

Evolutionary approach – where a series of prototypes are produced and examined by the users, feedback is received and the prototype amended, and this cycle of “creation and feedback” continues until the final prototype becomes the actual system to be used. The motto of this approach is “don’t expect to get things right first time”, as iteration is expected until the prototype is deem acceptable to satisfy the needs of the users. Normally, this approach is used with small scale systems that have dynamic or initially unclear user requirements.

Phased approach – similar to the traditional approach, except that the project is carved up into phases of development, so that each phase is suitably manageable, and can be delivered within a relatively short timescale. Users then gain benefits earlier than in the traditional approach. This approach is suitable when the project has defined aspects that can be separated and developed at different times. The overall requirements still need to be clear and well defined, so that phases can be identified and appropriately scheduled.

Overall user requirements for data within a data warehouse will probably be fairly well defined and the data structures will not change rapidly. It is also stated to be a large data warehouse. For these reasons, it is probably best that a phased approach is used (indeed, a phased approach is frequently used in practice e.g. in the case of the well-documented First American Corporation (FAC) data warehouse development) so that some parts of the data warehouse (i.e., the implementation of the support for 1 or 2 Key Performance Indicators (KPIs)) can be developed quickly and thus the benefits of the data warehouse can be felt quickly within aspects of the organisation: this would be much slower if a pure traditional approach was used. Indeed, a criticism often levelled at the traditional approach is that the time it takes to develop the entire data warehouse is very long and consequently management lose interest in the system (as it is not visibly producing anything of value for potentially several months). An evolutionary approach would not be as suitable, given the generally easy identification of the data requirements, and indeed may be quite difficult to manage in the case of such a large data warehouse.