R68-43 Two Continuous System Modelling Programs

The author has chosen a somewhat inappropriate title for this paper; the file records he discusses are not "generalized," but rather tailored to predictable real world constraints, and the "management information systems" he considers form only a portion of the spectrum. Nevertheless, he presents an objective technique for file record design quite applicable in certain cases. Given 1) known and predictable data and basic interrelationships subject to 2) known and predictable use (processing/querying), then the design techniques presented by the author should certainly ease the designer's conscience, if nothing else, and will probably materially assist in "tuning" a recordoriented data management system. The analytic design techniques presented form one element in a suggested iterative design loop "of data collection, preliminary file design, simulate (sic), analyze (sic) results...." Additional consideraton of "tuning" a system might well lead to consideration of special instrumentation, special reduction for the data gathered by such instrumentation, adaptive systems which "tune" themselves for optimum performance, and other subjects. This article concentrates on the analytic design techniques, to the exclusion of the other aspects of the complex design cycle. The author discusses techniques which require a rather large investment in manpower to apply, even if one assumes that "most designers will have a linear programming package available to them, or can have one written," as does Benner. The techniques presented have application to data/procedure sets in which large quantities of relatively predictable data are processed in a relatively predictable fashion. Since the techniques are strictly record-oriented, they are not directly applicable to some data management systems (e.g., TDMS,l ADAM2), which are object-oriented, or which use list structures. The "management" being serviced by systems amenable to Benner's techniques is more probably lower management, rather than middle or upper management-the information requirements of the latter groups tend to be less predictable than those of the former.