Size & Estimation of Data Warehouse Systems

Data Warehouse Systems are a special context for the application of functional software metrics. The use of an unique standard, as Function Point, gives serious comparability issues with traditional systems or other paradigms, in terms of both numerical size and implementation effort estimation. Peculiar guidelines are therefore necessary in order to identify the user view, the software boundaries, the data and the transactional components of such systems. Particularly, the boundary identification may strongly affect the measurement result for a data warehouse project; consequently, one can find huge, unacceptable deviations in the estimation of effort, time and cost for the given project. This paper shows the substantial differences between “traditional” software and data warehouse systems, the main guidelines that one can use when measuring the latter, and peculiar considerations for differentiating the effort estimation by measured element types. The depicted case studies highlight the fundamental relevance of the concept of “layer”, as explicitly stated by the most recent evolutions in the functional metrics field (COSMIC Full Function Point) in evaluating those functions which are seemingly transparent to the final user, but which cannot be neglected when estimating the implementation effort of the measured system.