Marketplace Microservice Platform

Event-driven marketplace backend built with Spring Boot, Kafka , Redis and ELK stack

Architecture overview

The system is built using microservice architecture and asyncronous communication via kafka

Services

|Service | Responsibilty | - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - | |auth-service | authentication, user events | |user-service | user profiles | |catalog-service | courses, tickets | |booking-service | course reservations | |payment-service | confirmation of booking ( simplified ) | |notification-service | email notifications | |analytic-service | kafka consumers , ELK (without Kibana) |

Booking & Payment Flow

1. User chooses a ticket ( frontend already should have list of AVAILABLE tickets of a certain catalog/course and probably it sends random available ticket binding it with userId )
2. booking-service receives request, creates in database booking despite status and redis set new variable with TTL 10 min (in redis cli "CONFIG SET notify-keyspace-events Ex" (without quotes) to automize expiration informing)
3. booking-service instantly sends request (through kafka) for payment to -> payment-service and -> catalog service to reserve it , it sends back to necessary topic (redis values deleted to prevent automated expired event) and booking-service sends request to catalog to take (change status SOLD)
4. payment status event also sent to notification-service which inform user about status of payment through email and stores details to its database

Kafka events

• UserEvent ->(topics = "users" , user.created)
• TicketEvent ->(topics = "ticket.fail","ticket.success","ticket.reserve" , the result of booking and payment)
• PaymentEvent ->(topics = "paymentevent", request for payment, topics = "payment.1","payment.2", (ordinal number of status) result of payment)

All inter-communication is asynchronous via kafka Synchronous communication is intentionally avoided to reduce coupling and improve resilience.

Tech stack

• Java 21
• Spring Boot
• Spring Kafka
• Redis
• PostgreSQL
• Kafka
• Elasticsearch + Logstash
• Docker & Docker-compose

How to run

docker compose up --build

Notes

This project uses simplified payment model. Real payment providers replaced with mock logic to focus on event-driven architecture

Full end-to-end testing is a work in progress End-to-end flow is partially tested; full integration testing is planned as a next step

Currently working on introducing a shared common-events module to enable event-driven chain flow and consistent communication across microservices

In the next iteration I plan to add contract tests and use analytics data to verify cross-service consistency