Can metrics and models be applied across multiple releases or projects?

Abstract

The issue of the applicability of metrics across multiple releases or projects is of great importance in measurement. If the same metrics can be applied across multiple releases or projects, significant cost savings can be realized in the management and implementation of the measurement program.For example, if one set of metrics can be identified on release n that are applicable on release n+l,..., N, the following cost savings would be achieved: Simplified management measurement program. of the Reduced metrics analysis time resulting from the fact that new sets of metrics do not have to be identified and validated for each release or project. Reduced training requirements because personnel do not have to be trained in several sets of metrics. Reduced data management cost because data about multiple sets of metrics do not have to be stored and retrieved.The panel will identify the types of applications and projects where this approach has been successful and where it has not, using product and process data from projects at Microsoft, Hewlett-Packard, Nortel, IE3M Global Services, and the Space Shuttle. The conditions that must exist for this approach to be successful, such as process maturity and stability, type of application, and types of product and process metrics, will also be covered.The panel will be stimulating and provocative because the panelists will take opposite positions on whether the approach is applicable to a variety of applications and projects or is restricted to special situations. Interaction with the audience will be a key element of the panel session. This will include the audience's experiences with the approach and their opinions about the question posed by the panel: Can Metrics and Models be Applied Across Multiple Releases or Projects?