A data complexity formula for deriving time-to-build estimates from non-relational to relational databases

Abstract

Despite the many qualitative elements of software time-to-build estimating, some observable features can be quantified; even if the resulting set of variables observed is arbitrary. Such is the case when estimating the expected duration for database re-engineering. If we assume that for any extant database, an entity-relationship model (ERM) can be produced from which a new normalised schema is generated, then our estimating task needs to quantify both the complexity of the ensuing ERM and also the data modelling knowledge of the ‘re-engineer’. Whilst there may be additional variables to be considered, a set of primary elements required for estimating the durations of the task have been identified. The formula proposed in this paper is arbitrary but it is intended as an instrument for measuring ER model complexity, such that time-to-build estimates can be made for the task of re-engineering extant non-relational databases into relational form.