Platform capability · Observability
OpenTelemetry, traces, logs and dashboards
Available for each deployment. Platform services send OpenTelemetry traces, metrics and structured logs through the OpenTelemetry Collector to Grafana, Jaeger, Loki and Prometheus. Dashboards cover infrastructure, dependencies, HTTP endpoints, request rate, errors, response time, the event stream and service-level error budgets. Alert rules in Grafana detect common operating failures, and each covered alert links to the runbook for responding to it.
Platform services get their OpenTelemetry setup from shared service code. Each production deployment supplies the Collector address and turns on the exporters it uses. Tracing of HTTP calls, database queries, messaging and caching is applied through that shared setup, so no service adds it separately.
A trace follows a request through HTTP calls and messages between instrumented services. Jaeger shows the connected steps of the request, so operators can see which service, dependency or background step caused an error or a delay.
The service-level dashboard shows availability, response time, a rolling 30-day error budget and how fast it is being used, for each service. Availability alert rules are generated from the documented target of each covered service and come with the dedicated monitoring stack. Paging destinations and proof of a test alert are completed for each deployment. Alerts on slow responses remain planned until the response-time measurements produce reliable percentiles.
Grafana rules separate warning and critical incidents, group related notifications and repeat unresolved alerts on separate schedules. Covered alerts include a summary, the affected service and a runbook link. Dedicated stacks send alerts through an authenticated Notifications endpoint that carries the customer and app instance. Where the customer receives them is configured for that deployment.
Operator runbooks cover service outages, high error rates, slow responses, OpenTelemetry Collector problems and unreachable dependencies. Each runbook names the alert and its effect, then gives the steps to investigate, fix and escalate using Grafana, Prometheus, Loki and Jaeger. A runbook for customer-managed key incidents is planned together with that feature.
A customer on dedicated infrastructure can have its own Grafana, Prometheus, Loki, Jaeger and OpenTelemetry Collector stack. The platform’s provisioning adds the customer and app instance to alert delivery, and Grafana sends alerts through the Notifications service to the operator channels you configure.
Additions that extend the capability described above.
A solution architect will demo the live service against your domain model.