Once we have conceptual data models, we go into the next stage, which is logical data modeling. In this stage, we do further analysis and scope of the data requirements. We will build out the details of the data models by asking questions or having discussions with business specialists. For example, you might have questions such as the following about the relationship between users and teams:

• How many teams can a user create?
• Does a user have to create a team?
• Besides creating a team, can a user join the teams created by other users?
• If so, how many teams can a user join?
• Can a user create a team with a name that is already being used by another team?
• How can we identify a team by its name?

With answers to questions like these, we start to gain a deeper understanding of the data requirements and the logic that haven't been revealed during the conceptual data modeling.

And, during logical data modeling, we will need to do normalization and usually denormalization. Normalization is a process for eliminating data redundancy and anomalies during data insertion, updating, and deletion and ensuring the data dependencies make sense. Denormalization is the opposite of normalization and is used mainly for performance improvement. We will also see an example of this later.

In both the conceptual data modeling stage and logical data modeling stage, usually, we will make generalizations with the use of subtypes and supertypes to adjust the data models. We will see an example of this later.