Observability is the ease of understanding an application’s runtime behavior and the behavior of its users. It provides feedback after deployment, helping teams detect failed releases, diagnose production issues, and understand feature usage. section 5.6 section 5.6.2
Observability depends on more than collecting telemetry. Logs, metrics, and distributed traces must also be stored, analyzed, and made accessible to developers. Analysis helps teams diagnose issues, understand dependencies, and assess reliability. section 5.1.3 section 5.6.2
Implement observability through shared responsibilities:
- Developers instrument components to emit telemetry. Libraries can provide some instrumentation, while domain-specific metrics and application-level logging may require custom code. section 5.6.2
- Operations provide infrastructure to collect, store, and analyze telemetry, including log aggregation and monitoring services. section 5.6.2
Deployment feedback
During deployment, a component emits error-rate and latency metrics. The observability infrastructure collects them and supplies them to the deployment mechanism, which determines whether the deployment succeeded or should be rolled back. section 5.6.1
Apply patterns such as log aggregation, distributed tracing, and application metrics to make behavior visible. For a large, complex application, a multi-component architecture can improve observability; a small monolith can also satisfy observability requirements. section 5.6.2 section 5.7 section 6.1.2
























