The OpenTelemetry project addresses the needs of modern, containerized environments running microservices and serverless applications on Kubernetes and cloud systems. This article contains a video showing how to integrate OpenTelemetry into a Java application using the Quarkus framework.
Scaling metrics collection
Today's scalable application platforms leave site reliability engineers (SREs) with the significant burden of collecting telemetry data ranging from a single cluster to multiple clouds. Contrast this challenge with traditional environments from years ago—you could probably get by with dashboards then, perhaps from third-party application performance monitoring (APM) tools.
Application topologies are getting more complex due to distributed, cloud-native microservices. Infrastructure environments also range from Linux containers to Internet of Things (IoT) edge devices, public clouds, and multiple Kubernetes installations. This diversity is also a new challenge for SRE teams and developers who want to trace application chains.
Benefits of OpenTelemetry
To solve the challenge, the OpenTelemetry project was created through a merger of the OpenTracing and OpenCensus projects. The key benefits of OpenTelemtry include:
- Create and ingest telemetry data from application services to IoT edge devices.
- Increase observability to monitor metrics, logging, and event tracing by a consistent collection mechanism.
Once the OpenTelemetry collector or client gathers telemetry data such as metrics, logs, and traces from your applications or app servers, you can send the metrics to a back-end server such as Jaeger to analyze applications' performance and behavior.
To take advantage of this powerful new platform, you have to install and configure the OpenTelemetry server itself, then change existing application projects to incorporate OpenTelemetry's capabilities. All this reworking has to take place on cloud-native microservices and serverless functions on Kubernetes.
Distributed tracing with Quarkus
The video in this article shows how Quarkus integrates the OpenTelemetry features for distributed tracing with serverless functions on Red Hat OpenShift. The OpenShift cluster enables developers to deploy cloud-native microservices as serverless functions using Knative serving. Find more information about Knative here.
The steps shown in the video include:
- On your local system: Implement multiple reactive, RESTful applications.
- On your local system: Collect local telemetry data by the OpenTelemetry collector in Quarkus Dev mode.
- On your local system: Visualize the telemetry data through the Jaeger console locally.
- On the remote Kubernetes system: Install the OpenTelemetry and Jaeger Operators using operators on Red Hat OpenShift.
- On the remote Kubernetes system: Deploy the reactive applications as serverless functions using Quarkus extensions and Knative on Red Hat OpenShift.
- On the remote Kubernetes system: Verify the telemetry data that OpenTelemetry collects instantly when the serverless functions get started and send the data to the Jaeger server.
Note: You can find the complete code for the example in my GitHub repository and replicate the tutorial in your development environment.
Continue your cloud-native journey with Quarkus
Here are more resources to get familiar with the ways Quarkus enables you to reduce your development time locally and jumpstart your development in Kubernetes without a lot of installation tasks: