WePlay — High-Performance Game Store Platform

- Performance Optimization (Redis) Implemented a multi-layer Redis caching strategy with a custom-tuned JedisPool (500 max connections), caching hot data including product details, user sessions, and order status. Applied TTL-based expiration, distributed locking to prevent cache breakdown, and randomized TTL offsets to guard against cache avalanche. Reduced average response latency from 394ms to 205ms (a 48% improvement). - Kafka-Based Async Architecture Deployed a 3-Broker Kafka cluster on a 96-core / 64GB host with replication factor of 3 and 10 partitions per topic. Controllers batch-produce messages to Kafka topics and spawn async threads for callback listening. Consumer services pull messages in batches and process them in parallel using multi-threaded workers, enabling traffic shaping and peak load absorption while keeping the database protected. - High-Load Stress Testing Conducted rigorous stress testing using Apache JMeter across incremental load phases (1,000 → 5,000 → 10,000 concurrent users). Validated system stability with a peak throughput of 477.3 requests/sec and sustained a 0.00% error rate across 10,000+ transaction samples. - Database Sharding & Partitioning Implemented horizontal sharding by user ID and order ID to distribute data across multiple database instances. Applied time-based RANGE partitioning on order tables and HASH partitioning on user tables to optimize query performance under large data volumes. - Stateless Authentication Built a JWT-based authentication system with Access/Refresh token rotation, bcrypt password hashing, and role-based access control, designed for horizontal scalability with no server-side session storage.