OpenTelemetry

OpenTelemetry is an open-source observability framework that provides vendor-agnostic instrumentation for collecting metrics, logs, and traces from your applications. Amazon CloudWatch supports OpenTelemetry natively across all three signal types: metrics queryable with PromQL for flexible analytics, logs searchable with Logs Insights, and traces explorable with Transaction Search. Because all three signals share a common data model with consistent attributes, you can correlate across metrics, logs, and traces to move from detection to root cause faster.

You can send custom OTel metrics directly to CloudWatch and query them using PromQL alongside AWS vended metrics. For the resources and metrics that support this, see Supported AWS resources and metrics for OpenTelemetry enrichment. PromQL gives you flexible and scalable analytics, so you can build dashboards, set CloudWatch Alarms, and explore metrics in Query Studio, the native PromQL console experience. OpenTelemetry metrics also carry rich semantic labels and support higher granularity than traditional CloudWatch metrics. This granularity enables precise filtering, aggregation, and analysis across your applications and infrastructure. OTel logs that you send to CloudWatch are also available in LiveTail for real-time streaming.

For application performance monitoring, CloudWatch Application Signals provides a unified, application-centric view of your services and dependencies. Application Signals uses OTLP traces to help you monitor application health, triage issues, and identify the impact on end users. With Transaction Search, you can explore OTLP spans interactively, find transactions using attributes such as customer identifiers or order numbers, correlate transactions to business events such as failed payments, and trace interactions between application components to establish root cause.