First dashboard in 10 minutes
Add a Prometheus data source, create one time-series panel for CPU usage, and save the dashboard in a folder named 'team-observability'.
DevOps & Tooling / Knowledge library
Grafana
open-source observability dashboards and alerting platform
Why learn Grafana?
Grafana gives teams one place to query, visualize, and alert on telemetry from many systems, reducing time-to-detect and speeding incident response.
Where it came from
Started in 2013 by Torkel Ödegaard, initially focused on Graphite-style monitoring dashboards. Grafana 1.0 was released in 2014 and adoption grew quickly in DevOps teams. Grafana Labs formed to maintain Grafana and build a larger observability ecosystem. In 2021, Grafana core licensing moved from Apache 2.0 to AGPLv3. Grafana 8 introduced unified alerting to reduce split alert workflows.
- Started in 2013 by Torkel Ödegaard, initially focused on Graphite-style monitoring dashboards.
- Grafana 1.0 was released in 2014 and adoption grew quickly in DevOps teams.
- Grafana Labs formed to maintain Grafana and build a larger observability ecosystem.
- In 2021, Grafana core licensing moved from Apache 2.0 to AGPLv3.
- Grafana 8 introduced unified alerting to reduce split alert workflows.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Add a Prometheus data source, create one time-series panel for CPU usage, and save the dashboard in a folder named 'team-observability'.
- Create a dashboard variable for service name, then build 3 panels (latency, error rate, request volume) that all use the variable.
- Create a Grafana managed alert: trigger when 5xx error rate is above 2% for 10m, route it to a contact point, and add runbook link annotations.
- Use provisioning files to auto-load one Prometheus data source and one dashboard on Grafana startup. Verify changes survive container restarts.
- From a high-latency panel, add links so users can jump to related Loki logs and Tempo traces for the same service and time range.
More ways to practice
Reusable multi-service dashboard
Create a dashboard variable for service name, then build 3 panels (latency, error rate, request volume) that all use the variable.
Actionable alerting
Create a Grafana managed alert: trigger when 5xx error rate is above 2% for 10m, route it to a contact point, and add runbook link annotations.
Provisioning-as-code
Use provisioning files to auto-load one Prometheus data source and one dashboard on Grafana startup. Verify changes survive container restarts.
Correlate metrics, logs, and traces
From a high-latency panel, add links so users can jump to related Loki logs and Tempo traces for the same service and time range.
Documentation and sources
- Grafana documentation (latest) (grafana.com)
- Provisioning (grafana.com)
- Alerting (grafana.com)
- Dashboard templates (grafana.com)
- Plugin development (grafana.com)
Catalog research notes
- Recent major releases improved dashboard rendering speed and navigation UX.
- AngularJS-based plugins are deprecated; modern plugins are React-first.
- Unified alerting has matured with better routing, silences, and contact points.
- Integrations continue to expand across Prometheus, Loki, Tempo, and OpenTelemetry.
Connect the ideas
Explore next
PrometheusOpenTelemetry (opentelemetry)KubernetesInfluxDBElasticsearchClickHouseOpenSearchTimescaleDB
Real World Example
Unified Logistics Telemetry Aggregation
A regional logistics firm managing a fleet of automated warehouse robots and delivery drones.
The business problem
Operational teams struggle with fragmented visibility, as robot diagnostic logs, drone flight telemetry, and warehouse environmental sensors reside in isolated silos, preventing rapid incident response.
How this topic solves it
Grafana acts as a centralized observability layer by utilizing its data source plugin architecture to query disparate systems—Prometheus for robot metrics, Loki for drone logs, and Postgres for warehouse sensor state—into a single pane of glass.
- Install Grafana OSS on a centralized management server.
- Configure individual data source plugins for Prometheus, Loki, and Postgres to establish connectivity.
- Create a unified dashboard using panel visualizations to correlate robot battery levels with drone flight logs.
- Define alerting rules within Grafana to trigger notifications when sensor thresholds are breached.
- Grant role-based access to regional managers for specific dashboard views.
This diagram illustrates how Grafana pulls data from various silos to provide a unified view.
Prometheus source
Loki source
Postgres source
Centralized Dashboard
End Users
- Robot Metrics → Query via Plugin → Grafana OSS
- Drone Logs → Query via Plugin → Grafana OSS
- Warehouse Sensors → Query via Plugin → Grafana OSS
- Grafana OSS → Visualize and Alert → Operations Team
Expected business outcome
Operational teams achieve a holistic view of the supply chain, enabling faster identification of cross-system failures and reducing the time spent manually correlating data across disconnected platforms.
Limitations and checks
The primary limitation is the potential for increased latency when querying high-cardinality data across multiple remote sources simultaneously, which can be checked by monitoring the query execution time in the Grafana Explore panel.