Grafana
Grafana is the open and composable observability platform. It queries, visualizes, and alerts on metrics, logs, and traces from many datasources in one place. We trial it as the default pane of glass for SRE and platform teams. Reach for it first when you want open standards and swap-friendly backends instead of a proprietary-only observability stack.
Blurb
Grafana allows you to query, visualize, alert on, and understand your metrics no matter where they are stored. Create, explore, and share beautiful dashboards with your team and foster a data-driven culture.
Summary
What it is: Open source dashboard and alerting stack (Grafana OSS) plus a large plugin ecosystem. Connects to Prometheus, Loki, Elasticsearch, InfluxDB, Postgres, cloud vendor APIs, and hundreds of other backends without forcing a single ingest pipeline.
When to use: Default observability UI to limit vendor lock-in: query in place, mix Prometheus, Loki, cloud APIs, and SQL without a mandatory ingest vendor. Unified monitoring for infra and apps; on-call alert routing; Explore workflows for metrics-to-logs drilldown.
When to skip: Primary need is governed warehouse BI with semantic layers and self-serve SQL for business users. Prefer Metabase, Apache Superset, or Redash for analyst-centric SQL dashboards. Pair Grafana with those tools rather than stretching it into full BI.
Key features: Panel plugins and templated variables, unified alerting, role-based access, correlations across signals, optional Grafana Cloud or enterprise extensions.
Details
| Topic | Notes |
|---|---|
| Deploy | Single binary, Docker, Helm, or Grafana Cloud; separate scale paths for query frontends vs alert workers |
| Data model | Metrics-first; SQL and logs via datasource plugins, not a built-in warehouse |
| Auth | OAuth, SAML, RBAC; map teams to folder and datasource permissions |
Practices: Prefer Grafana OSS or self-hosted deploys when lock-in risk matters; keep dashboard JSON in git; label alert routes by severity. Avoid using Grafana as the only BI layer for finance or product analytics without a semantic layer elsewhere.
References