Maintaining Consistency between Data Models and Databases: An Alternative Process

A. McAllister and J. Mattinson (Canada)


database design, conceptual data model, database tuning


Maintaining consistency between an entity-relationship model and the relational database it represents can be problematic. After a model is used to generate a database definition (often as SQL), a typical strategy for handling new requirements is to change the database definition directly. The data model can become out of date unless it is updated as well. Manual updates to the model are time consuming, are not guaranteed to be consistent with the database definition, and are commonly neglected in prac tice. Reverse engineering can be used to capture periodically an up-to-date model, however there are difficulties in obtaining accurate conceptual data requirements using automated reverse engineering tools. A new approach is proposed that ensures data models are up-to-date and re moves the need for reverse engineering. This approach uses tuning transformations as a concise method for specifying database design modifications related to performance tuning. Tuning transformations are represented separately, not embedding them in the ER model, nor in the database definition. This new approach addresses many of the problems discussed above for existing approaches. As a proof of concept, a prototype tool is described, which helps to ensure that an ER model remains consistent with the database definition.

Important Links:

Go Back