Performance oriented Spring Data JPA & Hibernate by Maciej Walkowiak

Turn off spring.jpa.open-in-view by default in Spring Boot applications to prevent holding database connections open unnecessarily

Change fetch type to LAZY for @ManyToOne and @OneToMany relationships instead of using default EAGER loading

Use projections instead of fetching full entities when you only need specific fields, especially for read operations

Monitor database connection usage with tools like Flexy, P6Spy, or DataSource-proxy to detect connection leaks and diagnose N+1 query problems

Avoid occupying database connections during external service calls by structuring transactions appropriately

Use @DynamicUpdate to only update changed columns instead of all columns when modifying entities

Consider using Spring’s TransactionTemplate for fine-grained transaction control instead of @Transactional with nested transactions

Implement proper connection pool sizing as the default of 10 connections can become a bottleneck

Use getReference() instead of findById() when you only need entity references without needing to load data

Be explicit about joins using @EntityGraph or JPQL join fetch statements to avoid N+1 queries

Consider using native SQL queries or JDBC for complex read operations where JPA mappings add unnecessary overhead