Howard M. Hess

Aligning technology and business: applying patterns for legacy transformation

Two key goals for aligning technology and business are to increase an organizations ability to change rapidly and to reduce the costs of technology. While many efforts are underway to improve application development, less emphasis has been placed on addressing key challenges posed by existing applications that resist rapid change. In this paper, we discuss techniques for accelerating change to legacy systems and for streamlining an application portfolio. Our approach takes business-driven application requirements, links them to analysis of an application portfolio, and identifies potential sequences of transformations to realize the targeted improvements. This paper describes our approach for mapping business requirements to application software, for using patterns to help translate business requirements to software requirements, and for using patterns to translate software requirements into potential solution designs. The paper describes how these techniques are applied to two stages of the software life cycle-initial analysis and detailed analysis-and summarizes experience gained from projects working with IBM clients.

Using logical data models for understanding and transforming legacy business applications

Many legacy applications perform essential business functions; yet, due to a number of factors, modifying such applications in order to accommodate new business requirements can be troublesome. Such factors include: the volume of code in a typical application, logical code structure that has deteriorated as updates have accumulated over time, functional redundancy, code structure that reflects the dated technology on which it was built, and scarce technical skills. We have argued that the consequent difficulty of understanding and modifying legacy code can be ameliorated through the use of logical data models. In the Mastery project, we are developing both algorithms for extracting logical data models from legacy COBOL applications and software tools that use the generated models to query and transform the code from which the models are derived.