Does Hibernate trigger in-memory pagination when using JOIN FETCH on collections?

Pagination with Collection Fetching

Hibernate utilizes setFirstResult() and setMaxResults() to implement pagination, typically translating these into dialect-specific SQL such as LIMIT and OFFSET. This process ensures that only a subset of records is retrieved from the database to optimize performance.

The Collection Join Conflict

A technical challenge arises when combining these pagination methods with JOIN FETCH on OneToMany associations. Because a join on a collection increases the number of rows returned for a single root entity, the database-level row count no longer aligns with the requested entity count.

It is understood that in certain scenarios, Hibernate may retrieve the entire result set into application memory to perform the pagination manually, which can lead to significant memory overhead for large datasets.

• Under what specific conditions does Hibernate decide to perform in-memory pagination instead of database-level pagination when a fetch join is present?
• Are there documented query strategies to paginate root entities while fetching associated collections without loading the full result set into memory?