code-optimization

Fix N+1 queries by replacing per-item repository calls with a single JOIN FETCH query, such as a @Query selecting orders with their items in one statement. This reduces N+1 queries to one and can cut P95 latency significantly.

Work from a performance analysis document and apply prioritized optimizations: better algorithms and data structures, query optimization, caching, and async processing. Measure each change with benchmarks and validate no functional regressions occur.

Yes, replacing a fixed thread pool with Executors.newVirtualThreadPerTaskExecutor lets I/O-bound tasks scale to thousands of concurrent operations. The documented example shows throughput increasing from 300 to 500 requests per second.

Use @Cacheable annotations with a CacheManager such as ConcurrentMapCacheManager for application-level caching, plus HTTP response caching headers and distributed caches like Redis for shared data. Always define cache invalidation strategies to maintain consistency.

Avoid optimizing without profiling data, since evidence-based changes prevent wasted effort. Never sacrifice functionality or readability for marginal gains, and apply one optimization at a time so each impact can be measured and rolled back.