In this project, I'll work on how to configure an API gateway to route traffic to our microservices architecture and implement solutions to improve the resilience of your system with patterns like circuit breakers, retries, fallbacks, and rate limiters using Spring Cloud Circuit Breaker and Resilience4J. Since the gateway is the entry point of your system, it’s also an excellent candidate to implement security concerns like user authentication. I'll work on how to do that with Spring Security, OAuth2, and OpenID Connect, relying on Spring Redis Reactive to manage sessions. Finally, I'll work on how to improve the observability of your system using Spring Boot Actuator and Spring Cloud Sleuth and relying on the Grafana stack.
You can use Docker Compose to set up the entire system, including applications, data services, and the Grafana observability stack.
First, package both the Edge Service and Book Service application as container images leveraging the Cloud Native Buildpacks integration provided by Spring Boot. For each application, run the following task:
./gradlew bootBuildImage
Then, from the project root folder, run Docker Compose.
docker-compose up -d
The Edge Service application is exposed on port 9000 while Book Service on port 9001. The applications require authentication through OAuth2/OpenID Connect. You can log in as Isabelle (isabelle/password) or Bjorn (bjorn/password).
Both Spring Boot applications are observable, as any cloud native application should. Prometheus metrics are exposed thanks to the Spring Boot Actuator library. Distributed tracing is enabled through Spring Cloud Sleuth.
Grafana lets us query and visualize logs, metrics, and traces from your applications. After running the Docker Compose configuration as explained in the previous section, we can access Grafana on port 3000. It provides already dashboards to visualize metrics from Spring Boot, Spring Cloud Gateway, and Spring Cloud Load Balancer. In the "Explore" panel, we can query logs from Loki, metrics from Prometheus, and traces from Tempo.
Fluent Bit collects logs from all containers and forwards them to Loki.
Loki is a log aggregation system part of the Grafana observability stack. "It's like Prometheus, but for logs." Logs are available for inspecting from Grafana.
Tempo is a distributed tracing backend part of the Grafana observability stack. Spring Boot applications sends traces to Tempo, which made them available for inspecting from Grafana.
Prometheus is a monitoring system part of the Grafana observability stack. It parses the metrics endpoints exposed by Spring Boot applications (/actuator/prometheus).
Metrics are available for inspecting and dashboarding from Grafana.