An estimation-based management framework for enhancive maintenance in commercial software products

Abstract

In commercial software vendor organizations, evolution of a software product by means of scheduled feature releases is a central activity. Relying on existing work on maintenance task effort prediction, this paper proposes a management framework for periodically capturing updated estimation data and using it as a basis for initial planning and subsequent re-planning of releases. The framework is founded upon a mathematically-stated, metrics-based model of the release cycle tuned to the software vendor environment. It allows non-technical product managers to make sensible, fast-paced and fine-grained release decisions without undue overhead or undue reliance upon the software development organization. This approach has been used successfully in practice by the author as vice president of the software development division of a mid-sized enterprise software vendor company.