Skip to content
Back to glossary
Observability

What is Grafana?

Grafana is the standard open-source platform for visualising metrics, logs and traces. Interactive dashboards, unified alerting, multi-source data.

What Grafana does

Grafana is the reference open-source visualisation and analytics tool in the observability ecosystem. Created in 2014 by Torkel Ödegaard, backed by Grafana Labs, it became the universal interface to explore monitoring data without forcing a specific backend. Crucially, Grafana does not store your data: it connects to Prometheus, Loki, Elasticsearch, PostgreSQL, ClickHouse, Tempo and another thirty sources, and builds dashboards that unify the view.

For a DevOps team, Grafana plays three distinct roles: explore metrics freely in ad-hoc mode (PromQL, LogQL, SQL depending on the source), assemble shareable dashboards for operational tracking, and trigger alerts based on thresholds or complex conditions.

The three pillars of observability in Grafana

Since version 8 (2021), Grafana has unified the three pillars of observability in a single interface:

  • Metrics via Prometheus, Mimir, Graphite or InfluxDB. Time-series view of counters, gauges, histograms.
  • Logs via Loki, Elasticsearch or Splunk. Text search, label filtering, correlation with metrics.
  • Traces via Tempo, Jaeger or Zipkin. Visualisation of distributed requests with their spans.

The operational benefit: during an incident, an on-call engineer clicks on a latency spike in a dashboard, jumps to the matching service logs, then follows the distributed trace to identify the root cause. All without switching tools.

Grafana vs proprietary solutions

Compared to Datadog, New Relic or Splunk, Grafana plays the portability card: no proprietary format, no lock-in, data stored wherever you want. That flexibility comes at a cost: you have to compose the stack yourself (Prometheus + Loki + Tempo + Grafana) rather than subscribing to an integrated platform.

For a Swiss SME, the typical trade-off is between 500 and 2,000 CHF per month (self-hosted open-source stack) versus 3,000 to 15,000 CHF per month (proprietary SaaS). For organisations subject to data sovereignty, self-hosting is often the only viable choice. Grafana Cloud (Grafana Labs' SaaS offering) provides a hybrid alternative with European hosting, better suited to nFADP/GDPR constraints.

Unified alerting

Since version 9, Grafana offers a unified alerting system that replaces notifications scattered between Prometheus Alertmanager, ElastAlert and other per-source tools. A single Grafana rule can evaluate several data sources simultaneously (for example, Prometheus latency AND Loki log error rate) and route to Slack, PagerDuty, Microsoft Teams, or a custom webhook.

On Hidora engagements, we systematically configure Alertmanager for native Prometheus alerts and Grafana alerting for multi-source correlations. Both coexist without conflict.

When Grafana is not the right fit

Grafana is designed for operational dashboards (10 to 100 concurrent users). For high-traffic dashboards (external customers, thousands of monthly views), it becomes heavy and costly to scale. In that case, solutions like Apache Superset or Metabase, dedicated to self-service analytics, are more appropriate.

Related Hidora services

  • Managed Services: deployment and 24/7 operation of a Grafana + Prometheus + Loki + Tempo stack, with standardised dashboards per client environment.
  • Consulting: observability strategy design, Grafana open-source vs Grafana Cloud selection, SLI/SLO definition on the dashboard side.
  • Prometheus, Observability, SRE: associated bricks in the modern observability chain.