Data Analysis and System Design by Entity-Relationship Modelling: A Practical Example

Entity-relationship modelling is valuable because it gives systems analysts a language for thinking in. The aim of this paper is to illustrate the thought processes inspired by the technique, by taking a practical example which contains a realistic level of complexity. The reader is forewarned that to follow the fine points of the discussion he will have to make a substantial investment of his time. Readers who are unfamiliar with the principles of system design through data analysis should read Ref. 2. A technical difficulty in systems analysis is the conception of an efficient overall system design, complete enough to fulfil all present requirements and robust enough to survive expansion and change. This difficulty is severe when the requirements are complex. The strength of the entity-relationship model lies in its ability to cut through the complexity, helping the analyst to explore alternative high-level decisions. A further value is the aid it gives to raising questions which should be put to management prior to concluding the design. An expository difficulty in explaining the use of the technique with a realistic example is the need to provide a rich set of facts describing the case, facts which the expositor should avoid marshalling in too orderly a fashion lest he destroys the point of the example. The facts which practising analysts acquire for background are usually discovered over a period of time, and arise in a disorderly fashion which is hard to simulate. This paper compromises with background facts from an old case whose design was not influenced by data analysis thinking. This also allows the process of dis-analysis to be illustrated. Dis-analysis is the construction of a data model from the record layouts of master and transaction files in a system. This is useful for interpreting system specifications for purposes such as tuning, estimation, maintenance and package evaluation. Presumably the analysts concerned with the original design faced an even richer set of facts than those recorded. I played no part in the case and have no special knowledge of the type of business, a bakery. Therefore I share with the reader the need to interpret the case purely at face value; the need to question the facts and to speculate about points of detail. The mention of limitations of the original design is certainly not meant as any slur on the analysts concerned, whose priorities and design philosophy must be judged in the context of the time. The reader should now study the facts in case study two, Order Control and Sales Accounting, Ref. 3.