---
title: "Track the health of your services | Grafana Labs"
description: "Service health, dependencies, and request metrics from OpenTelemetry data."
---

> For a curated documentation index, see [llms.txt](/llms.txt). For the complete documentation index, see [llms-full.txt](/llms-full.txt).

## Service-level visibility

[Application Observability](/docs/grafana-cloud/monitor-applications/application-observability/) is built natively on OpenTelemetry and lives inside the same Grafana Cloud platform as your other telemetry. It:

- Tells you when your services are slow or broken
- Helps you trace what went wrong across microservices
- Gets you to the root cause without switching tools

[Try it out in Grafana Play](https://play.grafana.org/a/grafana-app-observability-app/services?instrumentedFilter=all&sortFilterId=serviceName&var-environmentValue=production&from=now-30m&to=now&timezone=utc)

## What keeps you up at night?

Role / WorriesWhat you get with the app

**SRE, On-Call Engineer**

- When something breaks in the early morning hours, the first 30 minutes are spent just finding the right data across multiple tools
- By the time you have context, the damage is done

<!--THE END-->

- Click from a service in the map straight to traces, then logs, then profiles. No tool switching.
- Automatic correlation means no manual timestamp matching across systems.

**Developer**

- My endpoint is slow, but I can’t tell if it’s my code, a downstream service, or a database call underneath
- Tracing it end-to-end requires expertise I don’t have

<!--THE END-->

- Auto-generated service maps show exactly where latency is introduced across your dependencies
- Guided root cause analysis takes you from alert to likely source without needing to be an observability expert

**Platform, Observability Engineer**

- Proprietary agents installed across every service
- If we ever want to switch vendors, we have to re-instrument everything from scratch

<!--THE END-->

- Built natively on OpenTelemetry. No proprietary agents to install or manage.
- Your instrumentation is portable; it works with any OTel-compatible backend if you ever need to move

**Engineering Manager, VP Engineering**

- APM bill grows with every host we add
- Paying for multiple overlapping tools: APM here, logging there, tracing somewhere else

<!--THE END-->

- Usage-based pricing: you pay for telemetry, not per host or container, so cost tracks usage, not fleet size
- Metrics, logs, traces, and profiles in one platform. Fewer contracts, less duplication.