Implementation of the Traced Services

[moritzsternemann]

Implementation of the Traced Services

This PR introduces the Apodini HotROD (rides on demand) example, mirroring functionality of the official Jaeger HotROD example. The example is used to showcase the automated distributed tracing functionality of Apodini and consists of four distinct microservices:

graph TD
    A(frontend) --> B(customer)
    A --> C(driver)
    A --> D(route)

Loading

• frontend serves the user interface and provides the API to request a ride, retrieving information from the other services.
• customer mimics a database service the holds the directory of customers.
• driver is used to retrieve a list of the nearest drivers by mimicing a Redis interface.
• route returns the ETA for a given pickup and dropoff location.

For further information on service functionality, see the Jaeger HotROD Example README.

Tracing

When interacting with the service, request handling is traced using the OpenTelemetry Collector as the tracing backend. The collector forwards traces to the Jaeger instance that is exposed on port 16686.

The Jaeger user interface displays the collected traces. Each driver request in the HotROD user interface is displayed with a find trace link which opens the trace of the respective request in the Jaeger user interface.

Development

The docker-compose-jaeger.yml file can be used to run the OpenTelemetry collector and Jaeger locally using $ docker compose -f docker-compose-jaeger.yml up.

The docker-compose-development.yml file can be used to test the setup by building the web services locally and running the OpenTelemetry collector and Jaeger using $ docker compose -f docker-compose-development.yml up.

Related PRs

• Tracing implementation in Apodini#410

Reviewer Nudging

I suggest to review the services "bottom-up". Start with cross-service utils in the Utils folder. Then continue with the route, customer, and driver services. Lastly, check out the frontend service.