In a recent post to his site Jonathan Reinink has written up a guide to using dynamic (Eloquent) relationships in Laravel applications by making use of subquery functionality. In it, he shows how to make use of the selectSub method to select additional information in a single query versus having the overhead of custom, hard-coded relationships.

When building web apps that interact with a database, I always have two goals in mind: keep database queries to a minimum [and] keep memory usage to a minimum. These goals can have a drastic impact on the performance of your app.

Developers are typically pretty good at the first goal. We're aware of N+1 style problems, and use techniques like eager-loading to limit database queries. However, we're not always the best at the second goalkeeping memory usage down. In fact, we sometimes do more harm than good trying to reduce database queries at the expense of memory usage.

He starts off with the challenge he's trying to solve: gathering login information for users in a performant way. He includes the schema for the users and logins table and shows the code of how a normal relationship select might look to get login information for each user (creating an N+1 issue).

To help solve the issue, they try caching the last login information but realize they can do better - this is where subqueries come in. They provide an example of using the selectSub method to get the login information, mapping it to a macro for easier use and defining scopes. Finally, the tutorial shows how to use this method to select information via dynamic relationships. It also talks about lazy-loading issues and if the same thing could be accomplished with a "has one" relationship.