Most developers do this:
// Slow: Fetches entire entities, forces dirty checking List<Post> posts = entityManager.createQuery("select p from Post p", Post.class).getResultList(); High-performance code does this: high-performance java persistence book pdf
No PDF cheat sheet teaches you that—because it is an architectural pattern, not a Hibernate property. Every "High-Performance Java Persistence" summary tells you to use JOIN FETCH carefully. They warn about Cartesian products. Most developers do this: // Slow: Fetches entire
The high-performance secret? Instead of updating item.current_price , you append a bid to a separate bid_history table and calculate the price on the fly via a materialized view. You bypass the lock entirely. The high-performance secret
// Fast: Fetches only what you need, immutable, no persistence context overhead List<PostDTO> posts = entityManager.createQuery("select new com.dto.PostDTO(p.id, p.title) from Post p", PostDTO.class).getResultList(); Why is this faster than the book's PDF suggests? Because you remove the Entity Manager from the equation. No snapshots. No comparisons. Just data transfer. Vlad Mihalcea’s book is fantastic, but the concepts evolve faster than print. If you search for a static PDF, you freeze your knowledge in time.
But don't close the tab. Because the real high-performance persistence isn't about the file format. It is about three counter-intuitive truths that most developers learn too late. The search for the "PDF" usually starts after a developer realizes that Hibernate generated 500 queries for a single REST call. The knee-jerk reaction is to abandon ORMs entirely.
Imagine an auction system. Ten users bid on the same item. With @Version , nine users will get OptimisticLockException . You retry. The database churns. Performance collapses.