Titel:

Enriching RUP with key success factors for large-scale custom software development projects

(Marianne Heinemann, Bettina Duwe, Prof. Gregor Engels)

Session:

Anwendung von Vorgehensmodellen

Datum und Uhrzeit:

Montag, 3. Mai 2010 von 10:00 bis 10:40 Uhr

Beschreibung:

Capgemini sd&m has a long-standing experience in executing custom software development (CSD) projects with a special focus on large-scale and rightshore projects, where resources are distributed over different locations in Germany, Poland (nearshoring) or India (farshoring).

For many years, that experience was brought together in a proprietary life cycle model for CSD projects. However, this proprietary life cycle model was not based on an existing industrial standard and therefore was difficult to communicate to clients as well as to employees with other cultural background.

Besides the proprietary life cycle model, Capgemini sd&m has developed its software engineering methodology named QUASAR that focuses on the special requirements for Enterprise Architecture and Custom Software Development projects. With having enhanced QUASAR at fast pace in the last years, it became important to not only update the life cycle model, but to additionally integrate the software engineering methodology.

With the continuously increasing trend to software industrialization and thus the growing demand for standardization, the goal was to build on a standard life cycle model and integrate the key factors of our successful software engineering methodology QUASAR.

2 Approach

To unify the understanding of our software engineering processes and work products, we defined an ontology of relevant software engineering notions as the basis for all further actions.

The RUP disciplines have been validated, and we found a need to extend and detail them in order to reflect the special requirements of custom software development.

Next step was to validate the work products. A fine-grained artifact model has been developed that elaborates the artifact hierarchy based on a categorization for different artifact types. The result is a comprehensive catalogue of artifacts for all software engineering disciplines. We identified the critical artifacts and defined relevant in- and output artifacts for each discipline.

In parallel to the development of the artifact model, we commenced the life cycle model evolution. We identified the key success factors of our traditional approach and substantiated RUP to explicitly reflect them.

We elaborated our life cycle model based on the three dimensions results, time and tasks, the latter being reflected by the RUP concepts of phases, disciplines and activities. The results dimension was one elementary pillar within our traditional model: We work with a concept of stages that coarsely define, to which degree the system has been built. The explicit definition of development stages allows parallelized work and thus a quicker project execution.

The approach to build a development stage is iterative, where cycling through the development activities occurs by component and includes testing and software integration. The resulting step-by-step integration of software from early on is seen as a further success factor within our traditional life cycle model and as such has been transferred to the new model.

The bridging element between the life cycle model and the software engineering methodology is a new kind of milestone that is not foreseen by RUP: the discipline milestone. These milestones define the to-be status of critical artifacts at special points in time for each discipline and based on our development stages.
The resulting life cycle model is called “Quasar project incremental” and is meant to be our standard life cycle model for large-scale and rightshore projects.

3 Evaluation

“Quasar project incremental” combines the standard RUP approach with our traditional key success factors for custom software development. Furthermore, it integrates project management with our software engineering methodology hence providing a comprehensive model of interaction of those two levels within one project that practically helps to communicate.

In the meantime, “Quasar project incremental” has been deployed in several custom software development projects. Concrete samples and a critical evaluation will be given in the presentation.

Referenten:

	Dr. Marianne Heinemann ist Projektmanagerin bei der Capgemini sd&m AG und beschäftigt sich derzeit mit dem Management von Großprojekten, Vorgehensmodellen und Application Lifecycle Management. Zuvor hat sie bei Accenture als Beraterin, bei der Citibank als IT-Projektmanagerin und Leiterin der Abteilung „Business Analysis“ und bei Hewlett-Packard als EMEA-Programmmanagerin und Process Lead internationaler BPO-Projekte gearbeitet. Frau Heinemann ist Project Management Professional des PMI.
	Bettina Duwe ist Projektmanagerin bei der Capgemini sd&m AG und beschäftigt sich derzeit mit dem Management und Controlling von IT-Großprojekten, mit Schwerpunkt auf der Standardisierung und Industrialisierung der Software-Entwicklung. Sie hat langjährige Erfahrung in der Leitung von Großprojekten und besonderes Interesse an der Gestaltung effizienter Prozesse und Organisationsstrukturen. Zuvor war sie als Managerin bei Accenture und als Projektleiterin bei debis tätig.
	Prof. Dr. Gregor Engels ist Hochschullehrer für Informationssysteme an der Universität Paderborn. Er leitet das s-lab (Software Quality Lab) an der Universität Paderborn und ist Wissenschaftlicher Direktor von Capgemini sd&m Research in München.