The second goal of data modeling is to achieve minimal redundancy in our database. When we store the same data multiple times or store a piece of data that can be derived from other data, we bring in redundancy to our database. Most of the time, redundancy is bad. The extra space that is required to store the redundant data is not the main reason for us to get rid of redundancy. It is the extra effort and the potential inconsistency in the data that require our effort to achieve minimal redundancy in data models.

For example, for our User entities, we will have a first name attribute and last name attribute. And, if we added another attribute, say, initials, which can be calculated based on the first name and last name, we introduced redundant data, and every time we changed the names, we would also have to update the initials and if somehow, which could happen, we forgot to do that, there would be inconsistency in our database. And it should be avoided when we design the data models.

On the other hand, sometimes we need to add redundant data, mainly for the performance improvement that it will provide. However, during conceptual data modeling, we should get rid of redundancy first, then, explicitly only add those that can improve performance during logical data modeling.