How descartesresearch/teastore works
Project type and purpose
• This is TeaStore, a distributed micro-service reference and test application that emulates an online tea store.
• It comprises six core microservices (auth, image, persistence, recommender, registry, webui) plus supporting tooling for deployment, load generation, instrumentation (Kieker), and end-to-end testing.
File structure
• Top-level folders:
• services/… – each subfolder is one Java microservice, built with Maven, packaged as Docker images, exposing REST endpoints via servlets.
• interfaces/… – shared entity definitions (Category, Product, User, Order, etc.) used by all services.
• utilities/… – helper Docker images (database container, base Java/Tomcat image, Kieker probes, RabbitMQ integration).
• examples/… – deployment manifests (Docker Compose, Kubernetes YAML, Helm charts) and load-generator scripts (JMeter, Locust, custom HTTP).
• e2e-tests/… – Cypress end-to-end tests against running TeaStore.
• tools/… – shell scripts for building/testing.
README for additional insights
• TeaStore uses REST for inter-service communication and Netflix Ribbon for client-side load balancing.
• Services register with a central registry and discover each other dynamically.
• Instrumented variants send monitoring data (via Kieker, RabbitMQ) for performance analysis.
Detail
• Six microservice components plus the registry.
• Shared “interfaces” module as a UML package or note that entity DTOs are common.
• External dependencies: Ribbon client-side load balancer, MySQL, RabbitMQ.
• Which services use which utilities (e.g., Kieker probe library in the instrumented containers).