The Backbone Approach to Evolutionary Systems Development

While the introduction of the waterfall lifecycle model was a vast improvement over ad hoc process, there are inherent deficiencies in the waterfall model that have led to variants on the theme, including incremental, spiral, and evolutionary models. This paper addresses the planned evolution of large-scale systems from the design and build of smaller components based on an incomplete, yet end-to-end system, termed a backbone. The paper will discuss both engineering and business advantages of this approach to system development and also present some of the lessons learned in applying the backbone life-cycle model in two very large case studies. In the first case study, the system was an innovative product that would best be described as a “megasystem”. Thus, there was no history on which to build and the backbone approach was selected to provide early customer feedback regarding the viability of the system. In the second case study, the backbone approach was applied to a major upgrade of an existing system. The issue here was the inclusion of a “piggy-backed system” with the potential that the added hardware and software might interfere with the basic system operations. The backbone approach was used to demonstrate the noninterference of the two systems. Many advocates of various life-cycle models tend only to concentrate on the successful aspects of their approach. This paper will also discuss the two most significant challenges encountered, namely, (i) timely, efficient integration and test and (ii) tool and process support for configuration management of overlapping versions.