Have you identified bottlenecks during performance testing? Describe the analysis you performed when 200 users loaded the application and you observed response time issues, and explain the recommendations you gave to the client to improve application performance.

During the load test with 200 concurrent users, I first collected baseline metrics: CPU, memory, disk I/O, network latency, and database query times. I used a monitoring stack (Prometheus + Grafana) to visualize real‑time resource usage. The analysis revealed that the database layer was the primary bottleneck: slow queries, high lock contention, and insufficient indexing caused response times to spike above 5 seconds. The application server also showed high thread pool saturation, indicating insufficient worker threads. I recommended adding read replicas to distribute query load, optimizing slow queries and adding indexes, increasing JVM heap size, and scaling the application tier horizontally. Additionally, I suggested implementing caching for frequently accessed data and enabling connection pooling. After applying these changes, a retest showed a 60% reduction in average response time and improved stability under peak load.