It may seem like a duplication of data to manage a separate instance CI for a database CI when there is only one instance, but by using the relationship correctly, you are preparing for a future when there could be a cluster implemented.
This way you have people referencing the right type of CI from the beginning, and don't need to change code, lists, reference qualifiers, etc. Even if you 'carry' the overhead yourself to implement it this way, and maintain both CIs, end users only need to go to the right class, and don't need to worry about the other class.
