Supporting Iterative Development - Oracle

Iterative development methodologies typically consist of a series of rapidly developed prototypes. These prototypes are used to define the system requirements as the system is being developed. These methodologies are attractive because of their ability to show the customers something tangible as development is taking place. However, there are a few common pitfalls that occur during iterative development that undermine its effectiveness.

First, effective versioning is not always used. Creating multiple versions of an application allows certain features to be “frozen” while others are changed. It also allows different sections of the application to be in development while others are in test. Too often, one version of the application is used for every iteration of every feature, resulting in an end product that is not adequately flexible to handle changing needs (which was the alleged purpose of the iterative development).

Second, the prototypes are not always thrown away. Prototypes are developed to give the customer an idea of what the final product will look like; they should not be intended as the foundation of a finished product. Using them as a foundation will not yield the most stable and flexible system possible. When performing iterative development, treat the prototypes as temporary legacy systems.

Third, the development/test/production divisions are clouded. The methodology for iterative development must very clearly define the conditions that have to be met before an application version can be moved to the next stage. It may be best to keep the prototype development completely separate from the development of the full application.

Finally, unrealistic timelines are often set. The same deliverables that applied to the structured methodology apply to the iterative methodology. The fact that the application is being developed at an accelerated pace does not imply that the deliverables will be any quicker to generate.