Evolution of a Java Web E‑commerce System Architecture: From Single‑Server Deployment to Microservices and Distributed Infrastructure

Stage 0 – Introduction : The example builds a simple Java Web e‑commerce system to illustrate progressive architectural changes.

Stage 1 – Single‑Machine Site : All components (Tomcat/JBoss, JSP/Servlet, Spring, MyBatis, MySQL, etc.) run on one host, forming a basic monolithic system.

Stage 2 – Separate Application and Database Servers : The web server and database are split onto different machines, improving load capacity and fault tolerance.

Stage 3 – Application Server Cluster : Multiple application servers are added; traffic is distributed using load‑balancing techniques such as HTTP redirects, DNS round‑robin, reverse proxy (Nginx/Apache), IP‑level, and data‑link layer balancing. Algorithms like RR, WRR, LC, WRR‑LC, SH, DH, etc., are discussed.

Stage 4 – Session Consistency Solutions : Approaches include session sticky (IP hash), session replication, centralized session storage (database), and cookie‑based sessions, each with pros and cons.

Stage 5 – Read/Write Splitting and Master‑Slave Replication : Introduces read/write separation and MySQL master‑slave replication, or third‑party middleware (e.g., MyCat) to handle data consistency.

Stage 6 – Search Engine for Read‑Heavy Queries : Uses an inverted‑index search engine to accelerate fuzzy product searches, noting the added maintenance overhead.

Stage 7 – Caching Layer : Discusses page‑level, application‑level (Guava, Memcached), and database‑level (Redis) caching, emphasizing the need for consistent hashing in cache clusters.

Stage 8 – Database Sharding : Covers vertical sharding (separating user, product, transaction data) and horizontal sharding (splitting large tables across multiple databases), with challenges such as routing, primary‑key handling, and pagination.

Stage 9 – Application Decomposition : Moves toward microservices by splitting user, product, and transaction domains, then adopts SOA to centralize common services.

Stage 10 – Message Middleware Integration : Introduces language‑agnostic messaging (e.g., Dubbo with Zookeeper) to enable reliable inter‑service communication, load‑balancing transparency, and operational monitoring.

Conclusion : The evolution path is illustrative; real‑world architecture must be tailored to specific business needs and growth patterns.