I have a question --- do we need a logical data model for a conversion? Here are my thoughts. I believe the answer is yes if the conversion has any of the following characteristics:
- The target application is created in-house. This application will more than likely be enhanced in the future, so good definitions and understanding of the relationships would be required.
- If the target application is purchased, BUT we are adding extensions. Again, this application will more than likely be enhanced in the future, BUT what is more important is that it will be upgraded at some point in time. In addition, we will need to understand the changes that will need to take place in the upgrade.
The answer may be NO if you are converting one application to another application with no extensions or changes.
Logical data models define the relationships between data with definition. The definitions (entity and attribute) should be as close to business terminology as possible. I do have add in technical information in the data model, but in a notes or comments field.
A few recommendations when reviewing a logical data model could include:
- Is the logical data model modeled to Third Normal Form. Meaning - all the attributes in a table are dependent on the primary key and ONLY the primary key. There should be no duplication or redundancy in the attributes or the entities.
- Primary Key, Alternate Keys and Foreign Keys are all identified.
- All definitions make business sense.