seckill

a basic e-commerce seckill project built by Spring Boot for learning to optimize the server.

1. Use MVC architecture and domain-driven layered model;
2. Use Spring Boot, Spring MVC, MyBatis, Nginx, Guava, Redis and RocketMQ;
3. Use Nginx for reverse proxy and distributed expansion to handle static resources and Ajax requests respectively;
4. The three-level cache constructed by Redis, Guava and Nginx shared dic caches product information and improves the QPS of the product detail page greatly;
5. Use Redis to cache transaction verification and inventory information, and use RocketMQ transactional messages to perform asynchronous inventory deduction, which greatly increases the TPS of the order operation and also ensures the final consistency of inventory data;
6. Use the "token + token gate + queue" method to cut peaks, which greatly reduces the flow of the order interface; uses the verification code to smooth the flow, and allocates the peak from 1 second to 5 seconds; uses the token bucket algorithm to limit the flow, to limit the single-machine TPS, and to ensure the availability of services;
7. Deployed Nginx, Tomcat, MySQL, Redis and RocketMQ in CentOS 8 (2 core 4G) environment, and used JMeter for the stress test (3000 threads * 20 times), the result shows that product details function QPS can reach 5500+ and order operation TPS up to 6400+.