app-observability

What problem does it solve?

Consolidates application, frontend, and AI telemetry into a single observability workflow so teams can detect, diagnose, and correlate rate/error/duration (RED) metrics, traces, logs, and profiles without fragmented toolchains or manual instrumentation guesswork.

Core Features & Use Cases

• Span-metrics based APM: Generates RED metrics from OpenTelemetry traces (via Tempo metrics-generator or OTel Collector spanmetrics) for service inventory, service overview, and operations-level latency analysis.
• Frontend Observability with Faro: Capture web vitals, session replay, errors, and frontend traces to correlate browser sessions with backend traces.
• AI/LLM Monitoring: Instrument LLM calls with OpenLIT and OTel GenAI conventions to track token usage, latency, cost, and evaluation metrics.
• Full-stack correlation: One-click navigation between metrics, exemplar traces, logs (Loki), and profiles (Pyroscope) and service maps using required resource attributes such as service.name, service.namespace, and deployment.environment.
• Use Case: Instrument a microservice frontend and backend to find p95 latency spikes, open exemplar traces in Tempo, replay affected user sessions from Faro, and inspect correlated logs and profiles.

Quick Start

Use the app-observability skill to instrument your services with OpenTelemetry, configure Alloy or direct OTLP export, enable Faro for the frontend and OpenLIT for AI telemetry, and verify RED metrics and trace-log-profile correlation in Grafana Cloud.