Observability Service

Observability Service

Dubbo internally maintains multiple dimensions of observable metrics and supports various ways of visual monitoring. Observable metrics are generally divided into three measurement dimensions:

  • Admin The Admin console visually displays applications, services, instances, and dependencies in the cluster. It supports the issuance of traffic governance rules and provides tools such as service testing, mock, and document management to improve R&D testing efficiency.

  • Metrics Dubbo collects a series of traffic metrics such as QPS, RT, number of successful requests, number of failed requests, etc., as well as a series of internal component statuses such as thread pool count, service health status, etc.

  • Tracing Dubbo is adapted to mainstream link tracing works in the industry, including Skywalking, Zipkin, and Jaeger, all of which support Dubbo service link tracing.

  • Logging Dubbo supports multiple log framework adaptations. For example, in the Java ecosystem, it supports Slf4j, Log4j2, Log4j, Logback, Jcl, etc. Users can choose the appropriate framework based on business needs. Dubbo also supports Access Log to record request traces.

Admin

The Admin console visually displays applications, services, instances, and dependencies in the cluster. It supports the issuance of traffic governance rules and provides tools such as service testing, mock, and document management to improve R&D testing efficiency.

Admin Effect

Metrics

Dubbo runtime collects core service metrics including qps, rt, total calls, successful calls, failed calls, and failure reasons statistics. To better monitor the service running status, Dubbo also provides monitoring of core component statuses such as thread pool count and service health status.

Metrics can be visualized through Grafana.

Grafana Effect

Tracing

Full link tracing is of great value for monitoring the running status of distributed systems. Dubbo implements runtime request tracking through Filter interceptors. By exporting tracking data to some mainstream implementations such as Zipkin, Skywalking, Jaeger, etc., full link tracking data analysis and visualization can be achieved.

Tracing Effect

In the future, we plan to support dynamically adjusting Dubbo’s link tracing sampling rate through governance platforms such as Dubbo Admin.

Logging

Access logs can help analyze the system’s traffic situation. In some scenarios, enabling access logs is also very helpful for troubleshooting issues.

Last modified September 30, 2024: Translate (badad1b8b1)