Semantic Normal Form for Performance-Critical Databases

Semantic Normal Form (SNF) compliant software is software that is immune to malignant change. SNF software extends and adapts to business needs without recourse to database restructuring or the rewriting of associated software. This stability eliminates most software maintenance costs while also enabling Just-in-Time systems implementation. SNF software is achieved largely because it embodies an SNF compliant conceptual database schema. SNF compliant database schemas consist of tables with a simple uniform structure; however, SNF schemas tend to have many more tables than non-SNF compliant versions. This means that SNF compliant schemas necessitate the frequent use of multiple query joins in data manipulation. This raises the question about the technical feasibility of building high-performance database transactions for SNF compliant systems. In this paper, we measure the performance premium for SNF compliant software by comparing the performance of an SNF system with a non-SNF version. The SNF version was indeed a little slower but a deeper analysis of both the comparison results and eld experience with industrial SNF systems suggests that not only is it possible to achieve high-performance SNF compliant systems, but that SNF compliance is particularly desirable for large complex performance critical databases.