Grafana
The default open-source dashboarding and visualization layer — it turns metrics, logs, and traces from many sources into graphs and alerts. It stores nothing itself.
What is Grafana?
Grafana is the de-facto open-source visualization and dashboarding platform. It connects to data sources — Prometheus for metrics, Loki for logs, Elasticsearch, Tempo/Jaeger for traces, SQL databases, cloud monitoring APIs — and renders them as dashboards: graphs, heatmaps, tables, gauges, and alert rules. The key thing to understand is that Grafana stores nothing: it’s a query-and-display layer that sits on top of whatever databases actually hold your data. It’s the “single pane of glass” most teams put in front of their observability stack.
Grafana + Prometheus, and the boundary between them
The canonical pairing is Grafana + Prometheus: Prometheus scrapes and stores metrics, Grafana queries and visualizes them. Understanding the division of labor prevents the most common confusion — Prometheus is the database and query engine; Grafana is the eyes. Grafana’s power is that it’s source-agnostic: one dashboard can combine a metrics panel from Prometheus, a logs panel from Loki, and a traces panel from Tempo, giving a unified view across all three pillars of observability without moving data between systems. This is why Grafana became ubiquitous — it doesn’t compete with your data stores, it unifies them.
The Grafana ecosystem and where costs hide
“Grafana” today spans more than dashboards. Grafana Labs ships Loki (logs), Tempo (traces), Mimir (scalable long-term Prometheus storage), and Grafana Cloud (their hosted stack) — a full observability suite built around the visualization core. That’s worth knowing when evaluating: adopting Grafana can mean adopting an ecosystem. On cost: self-hosted Grafana is free, but the storage behind it (long-retention metrics, high-volume logs) is where spend accumulates, and managed observability pricing often scales with data volume and cardinality — the bill tends to come from the data layer, not Grafana itself. Alerting lives in Grafana too, overlapping with Prometheus’s Alertmanager; teams should pick one alerting path to avoid duplicate, conflicting alerts.
What people get wrong
- Thinking Grafana stores data — it’s a visualization layer; the metrics/logs live in Prometheus, Loki, etc.
- Duplicating alerting: running both Grafana alerts and Alertmanager for the same signals creates noise and confusion.
- Underestimating data-layer cost — Grafana is free, but the storage volume behind rich dashboards is what drives the bill.
Primary source: Grafana documentation
Historical figures and technical concepts for informational purposes only. Not technical, professional, legal, or financial advice. Sources: Official Documentation.