congiuluc/performance-backend-java

Backend Performance Optimization (Java)

When to Use

• Adding caching to frequently accessed data (Spring Cache, Redis, Caffeine)
• Optimizing JPA/Hibernate queries or repository methods
• Implementing pagination for large datasets
• Adding response compression (Gzip, Brotli)
• Profiling and fixing Java backend performance bottlenecks
• Tuning JVM settings for throughput or latency

Official Documentation

• Spring Cache Abstraction
• Spring Data JPA — Pagination
• Hibernate Performance Tuning
• Spring Boot Actuator
• Micrometer

Procedure

1. Identify bottleneck: database query, I/O wait, serialization, missing index, JVM GC
2. Apply caching and optimization patterns
3. Review caching sample
4. Add pagination for list endpoints (Spring Data Pageable, cursor-based)
5. Use @EntityGraph or JOIN FETCH to avoid N+1 queries
6. Use projections / DTOs instead of loading full entities for read-only use cases
7. Measure before and after with Micrometer metrics and Actuator
8. Add performance-sensitive tests for critical paths

Performance Review Checklist

| # | Check | Tool / Method | |---|-------|---------------| | 1 | All list endpoints accept Pageable | Code review | | 2 | Read-only queries use @Transactional(readOnly = true) | Code review | | 3 | Only needed columns projected (DTO projections) | Code review | | 4 | No N+1 queries (use @EntityGraph or batch fetching) | Hibernate SQL logging | | 5 | Response compression enabled (Gzip) | server.compression.enabled=true | | 6 | Hot data cached with @Cacheable and appropriate TTL | Code review | | 7 | Cache invalidation with @CacheEvict on write ops | Code review | | 8 | No blocking I/O in reactive/virtual-thread context | Code review | | 9 | Connection pool sized correctly (HikariCP) | Config review | | 10 | Database indexes on frequently queried columns | Schema review | | 11 | No EAGER fetch types on @ManyToOne/@OneToMany | Code review | | 12 | Batch inserts/updates use JDBC batching | Hibernate config review |

Profiling Workflow

1. Enable Hibernate SQL logging: spring.jpa.show-sql=true + logging.level.org.hibernate.SQL=DEBUG
2. Use Spring Boot Actuator metrics: /actuator/metrics/http.server.requests
3. Use jconsole or VisualVM for JVM profiling
4. Use async-profiler for CPU flame graphs
5. Use Micrometer timers for custom operation timing
6. Run JMH benchmarks for micro-benchmarks on critical paths