Instrumentation provides application and platform code paths that create OpenTelemetry data. It can be added automatically through agents, framework modules, annotations, logging integrations, Kubernetes injection, or monitoring agents; SDKs also support custom signals. section 3.6.7
Choose the level that fits the signal source. Language-library instrumentation can use libraries and agent logic; framework modules can inject OpenTelemetry invocations into application engines; logging plugins can forward configured log events to an OTLP collector. Kubernetes configuration can inject instrumentation for commonly used runtimes. section 3.6.7
Automated instrumentation reduces manual changes, while explicit calls add business-aware boundaries:
- Use automated code instrumentation to establish OpenTelemetry coverage across application code and supported layers. section 11.3.1
- Add explicit OpenTelemetry calls when a span must represent a business transaction, identifying its start and end. section 11.3.1
- Configure logging-framework integrations, such as Log4j2 Appender or Go Zap plugins, when log events should be sent to an OTLP collector. section 3.6.7
Business-transaction spans
A team can treat automated instrumentation as an intermediate step, then add explicit calls around a business transaction so its span has a defined beginning and end. This complements a move toward OpenTelemetry-standardized events and OTLP collection. section 11.3.1
