OpenTelemetry统一可观测性：Traces、Metrics与Logs
OpenTelemetry Unified Observability: Traces, Metrics, and Logs
| iDev Tech | 2026-08-24T14:00:00
全面讲解OpenTelemetry如何统一分布式追踪、指标和日志三大可观测性信号。
Comprehensive guide on how OpenTelemetry unifies distributed tracing, metrics, and logs — the three observability signals.
为什么选择OpenTelemetryOpenTelemetry（OTel）是CNCF的毕业项目，也是可观测性领域的事实标准。它提供统一的SDK和API来采集Traces、Metrics和Logs三种信号，消除了供应商锁定——你可以自由切换Jaeger、Zipkin、Prometheus、Grafana等后端。Traces：分布式追踪Trace由多个Span组成，记录一个请求在分布式系统中的完整调用链。Spring Boot应用通过spring-boot-starter-actuator + opentelemetry-javaagent自动埋点。关键配置：Service Name、Sampling Rate、Exporter（OTLP→Jaeger）。实战集成在Spring Boot项目中集成OTel的步骤：添加OpenTelemetry Java Agent作为-javaagent参数启动；配置OTEL_EXPORTER_OTLP_ENDPOINT指向收集器；自定义Span（使用@WithSpan注解或手动创建）；通过Baggage传递跨服务的上下文信息（如用户ID、租户ID）。
Why OpenTelemetryOpenTelemetry (OTel) is a CNCF graduated project and the de facto standard for observability. It provides unified SDKs and APIs for collecting Traces, Metrics, and Logs, eliminating vendor lock-in — freely switch between Jaeger, Zipkin, Prometheus, Grafana, and other backends.Traces: Distributed TracingA Trace consists of multiple Spans recording a request's complete call chain through a distributed system. Spring Boot applications auto-instrument via spring-boot-starter-actuator + opentelemetry-javaagent. Key configuration: Service Name, Sampling Rate, Exporter (OTLP→Jaeger).Practical IntegrationSteps to integrate OTel in Spring Boot: add the OpenTelemetry Java Agent as a -javaagent startup parameter; configure OTEL_EXPORTER_OTLP_ENDPOINT to point to the collector; customize Spans (using @WithSpan annotation or manual creation); pass cross-service context via Baggage (user ID, tenant ID, etc.).