Arthur Fuller
fuller.artful at gmail.com
Wed Mar 12 10:06:14 CDT 2014
Mark et. al. In terms of data-modeling, this describes the difference between Logical and Physical data models. Logical describes a database in terms closer to human than to machine. Physical describes the actual implementation, taking into account such features as Sequences (Oracle for example). Logical describes the domain without regard for which RDMS one ultimately chooses. Tools such as ERWin, PowerDesigner, etc. exist because of this division. At the Logical level, the focus is on the problem or challenge or domain, and good tools free you from commitment to any particular RDMS. We do our design at the LDM level, and only when that level is satisfied (i.e. fully descriptive) do we move to the PDM level, whose choice of RDMS may have absolutely nothing to do with the particular problem (i.e. "We're an Oracle shop and that's that."). I've been in the db game for 3+ decades, and I deem data-modeling tools absolutely vital, even for relatively small apps consisting of say 20 tables or fewer. I've also, in fact more frequently, been on the farther end of the spectrum, designing systems involving 500+ tables; in those situations, you'd have to be either insane or masochistic or some combination of both to try it by hand. A.