Research on Frontend-Backend Collaboration and Performance Optimization for High-Concurrency Web Systems

Modern web applications commonly face imbalances between frontend and backend rendering and inefficient resource utilization in high-concurrency scenarios. This paper proposes a full-stack collaborative optimization framework integrating React Server Components with Node.js load-tiered scheduling. This approach first enables dynamic switching between frontend SSR and CSR through Async Render Tree Partitioning. It then introduces Nginx reverse proxy and multi-level Redis caching to reduce database access latency. Finally, it employs an event-loop-based Adaptive Task Scheduler on the backend to dynamically allocate I/O load. Experiments conducted in K6 and JMeter simulation environments demonstrate that under 10,000 concurrent requests, average response time decreased by 38.6%, P95 latency dropped by 41%, and server energy consumption reduced by 18.3%. This research provides a systematic engineering solution for performance optimization in large-scale web systems.