JPA / Hibernate – Architect-Level Interview Q&A

 

 1. What is the difference between JPA and Hibernate?

Answer:

• JPA (Java Persistence API) is a specification.

• Hibernate is a JPA implementation (and more) that provides additional features like caching, custom queries, and batch processing.

• JPA defines annotations, EntityManager, and query language (JPQL); Hibernate provides the actual engine and can be used in native (non-JPA) mode.

---

🏗 2. How do you manage entity relationships in a large domain model?

Answer:

• Use proper annotations: @OneToOne, @OneToMany, @ManyToOne, @ManyToMany.

• Avoid bidirectional relationships unless necessary to prevent infinite recursion and cyclic dependencies.

• Use @JsonManagedReference / @JsonBackReference for serialization issues.

• Prefer LAZY fetching for associations unless eager loading is necessary.

---

📊 3. How do you optimize performance in Hibernate-heavy applications?

Answer:

• Batching: Use hibernate.jdbc.batch_size to reduce round trips.

• Fetch Strategy: Use LAZY vs EAGER wisely. Avoid n+1 problems with JOIN FETCH or @EntityGraph.

• Caching:

  • 1st Level: Built-in per session.

  • 2nd Level: Use EhCache, Redis, or Hazelcast.

  • Query Cache: Enable carefully with selective query regions.

• Pagination: Always use pagination for large result sets.

• DTO Projection: Use Constructor expressions, native SQL, or Spring Data projections for read-heavy use cases.

---

🧠 4. Explain the Hibernate Session Lifecycle.

Answer:

1. Transient: New object, not associated with DB.

2. Persistent: Associated with a Hibernate Session.

3. Detached: After session is closed, but object still exists.

4. Removed: Marked for deletion.

---

⚙️ 5. What is the N+1 problem and how do you fix it?

Answer:
Occurs when a single query loads multiple parent entities, but each associated child is fetched in a separate query.

Fixes:

• Use @EntityGraph or JOIN FETCH.

• Batch size configuration: hibernate.default_batch_fetch_size.

• Use DTO-based projections.

---

🔁 6. How do you handle circular dependencies in bi-directional mappings?

Answer:

• Mark one side as @JsonManagedReference and the other as @JsonBackReference if using Jackson.

• Or use @JsonIgnore on one side.

• Alternatively, use DTOs to break dependency for serialization.

---

🛑 7. What are some pitfalls with Cascade operations?

Answer:

• Overuse of CascadeType.ALL may lead to unintended deletions.

• Avoid cascading REMOVE in collections unless strictly needed.

• Be explicit: prefer PERSIST or MERGE only.

---

🧩 8. How do you handle inheritance in JPA?

Answer:

• SINGLE_TABLE: One table for all types (use discriminator column).

• JOINED: Parent/child mapped to separate tables with join.

• TABLE_PER_CLASS: Each subclass gets its own table (less common).

Prefer JOINED for normalized DBs, or SINGLE_TABLE for performance.

---

💾 9. How do you handle multi-tenancy in Hibernate?

Answer:

• Use Hibernate Multi-Tenant support:

  • Schema-based: One DB, multiple schemas.

  • Database-based: Separate DBs per tenant.

  • Discriminator-based: One table, add tenant_id.

• Implement a CurrentTenantIdentifierResolver and MultiTenantConnectionProvider.

---

🛠 10. How do you tune Hibernate for large-scale applications?

Answer:

• Enable second-level cache selectively.

• Use stateless sessions for bulk inserts.

• Avoid over-fetching by limiting graph depth.

• Optimize entity mappings (avoid @ManyToMany where possible).

• Use connection pooling (HikariCP).

• Enable slow query logging.

---

🧪 11. How do you structure repositories in a large modular project?

Answer:

• Use Spring Data JPA interfaces with custom base interfaces (BaseRepository<T, ID>).

• Implement custom queries with @Query or NamedQuery.

• Group repositories by domain or bounded context.

• Avoid business logic in repositories—delegate to service layer.

---

🔍 12. How do you debug performance issues in JPA/Hibernate?

Answer:

• Enable SQL logging: spring.jpa.show-sql=true, Hibernate format SQL.

• Use tools: JProfiler, YourKit, or Hibernate Statistics.

• Track slow queries using logs or DB monitoring.

• Profile batch size issues or n+1 queries.

---

🔧 13. What’s the difference between merge() and saveOrUpdate()?

Answer:

• saveOrUpdate(): Saves a transient or detached instance.

• merge(): Returns a new managed copy; original remains detached.

• Prefer merge() when working with detached entities from outside session context.

---

💬 14. How do you enforce audit tracking in entities?

Answer:

• Use @CreatedDate, @LastModifiedDate with @EnableJpaAuditing.

• Store created/modified by with Spring Security context.

• Alternatively, use Hibernate Envers for historical tracking.

---

📦 15. Should you map everything as entities?

Answer:
No. Use entities for write-heavy models. Use DTOs, views, or native queries for:

• Read-heavy or reporting use cases.

• Aggregated or denormalized data.

• Reducing entity complexity.