A developer outlines a four-service microservices architecture for an e-commerce platform. The system uses Django for most services, with a dedicated API Clients for centralized authentication via tokens. Each service runs in its own Docker container, has an independent database (MariaDB or Redis), and is deployed separately on Kubernetes. This separation allows for independent scaling and development cycles.

The API Catalogue manages products and reviews, using Django signals to automatically update average ratings. The API Panier uses Flask and Redis for temporary cart storage, prioritizing speed over persistence. The API Commandes orchestrates workflows, validating stock and creating order snapshots to preserve pricing at the moment of purchase, a key practice for audit trails.

Key pitfalls include managing inter-service communication. The author stresses the necessity of adding timeouts to all HTTP requests and implementing robust error handling to prevent cascading failures. A critical architectural decision was enforcing unidirectional communication to avoid circular dependencies. This design trades the simplicity of a monolith for better scalability and clearer service boundaries.