Grafana

Visit Website

Introduction

Grafana is a dashboard and alerting server. It queries data sources and draws panels. It does not store your metrics by itself. Typical companions from the same organization are Prometheus or Mimir for metrics, Loki for logs, and Tempo for traces. You can also point Grafana at Elasticsearch, CloudWatch, BigQuery, and a long plugin list. The open-source server is grafana/grafana, licensed AGPL-3.0.

Grafana Cloud is Grafana Labs’ hosted version of that stack: Grafana, metrics, logs, traces, profiles, and k6 performance tests, with usage limits. The cloud product is paid after the free tier. Self-hosting the AGPL server is a separate choice from buying Cloud. Grafana is an observability tool. Testers use it to watch error rates and test pipelines, not to click through a UI.

Features

1. Panels and dashboards: Time series, tables, logs, and traces in one board. Variables template a dashboard across services.
2. Data sources: Prometheus, Loki, Tempo, Elasticsearch, SQL databases, and cloud vendor APIs. A panel’s query language is the data source’s language, not a Grafana-specific one.
3. Alerting: Alert rules can live in Grafana and route to email, Slack, PagerDuty, and other contact points. You still need a query that means something.
4. Explore: An ad-hoc query view for a data source, separate from the curated dashboard.
5. Plugins: Official and community plugins add panels and data sources. Check who maintains a plugin before you depend on it.
6. Organizations and folders: Dashboards can be split by team, with view and edit roles.
7. Grafana Cloud integrations: Kubernetes, cloud accounts, and application receivers ship telemetry into the hosted backends without you running Mimir and Loki.
8. k6 and frontend tools: Grafana Labs also ships k6 and Faro. They show up in Cloud pricing because their telemetry lands in the same account. They are separate tools.

Pros

1. The server is free to run: AGPL-3.0 Grafana OSS is the UI most Prometheus installs already use.
2. One UI for several backends: Metrics, logs, and traces can be linked with exemplars and trace ids if you set the data sources up that way.
3. Cloud has a real free tier: Grafana’s pricing page lists an always-free Cloud plan with every Cloud service, usage limits, community support, and 14-day retention.
4. Dashboards as code: JSON dashboards and Terraform or Grizzly provisioning keep boards in git.
5. Hiring pool: A lot of engineers have already used Grafana. You are not introducing a private query UI.

Cons

1. Grafana is not the database: A pretty empty dashboard means Prometheus, Loki, or an agent was never installed. Budget the backend, not only the UI.
2. AGPL obligations: If you modify Grafana and offer it as a service, AGPL duties apply. Using unmodified Grafana internally is the common case and is not the same as selling a hosted fork.
3. Cloud bills follow usage: Pro starts at $19 per month plus what you ingest past the free tier. Logs and traces are the usual surprise, same as any hosted backend.
4. Alerting sprawl: Every team copies a dashboard and a slightly different alert. Without owners, pages fire for panels nobody watches.
5. Not a test runner: Grafana will graph a pass rate you export. It will not execute Playwright or talk to a device.

Usage and Links (Example)

Local OSS server with Docker:

docker run --rm -p 3000:3000 grafana/grafana

Open http://localhost:3000 (default login admin / admin, which you change immediately). Add a Prometheus data source and import a dashboard. That container has no metrics until something scrapes a target.

For Grafana Cloud, create an account, install the Kubernetes or host integration the stack wizard gives you, and send data to the provided remote-write URL and token. Keep the token in a secret.

Resources:

• Grafana website
• Grafana OSS docs
• Pricing
• GitHub: grafana/grafana

Pricing

1. Grafana OSS: Free to self-host under AGPL-3.0. You operate the server and whatever databases sit behind it. Grafana Enterprise (self-managed features such as reporting and extended auth) is a paid license on top of OSS; ask Grafana Labs for that number.
2. Grafana Cloud Free: $0, described as always free. All Cloud services with usage limits, 14-day retention for metrics, logs, traces, profiles, and k6 results, and community support.
3. Cloud Pro: From $19 per month plus usage. Longer retention (13 months for metrics, 30 days for logs, traces, profiles, and k6 on the published plan) and email support.
4. Cloud Enterprise: Starts at a $25,000 per year spend commit. Custom retention, premium support, and deployment options including public cloud and bring-your-own cloud.

The free Cloud tier is for evaluation and small projects. A production log pipeline will leave it.

Recommended for?

1. Teams running Prometheus who need dashboards and alerts without buying a UI.
2. Platform groups standardizing on Loki and Tempo as well as metrics, and willing to run those backends or pay for Cloud.
3. Small projects that can live inside Cloud’s free limits and 14-day retention.
4. Test leads who want pass-rate or error-budget boards fed by CI and production metrics. The tests still run somewhere else.

Conclusion

Grafana is the dashboard. Decide separately whether you will run the backends or pay Grafana Cloud. OSS is the right start when Prometheus is already in the cluster. Cloud’s free tier is enough to learn the product; Pro at $19 plus usage is the point where retention and ingest start to look like a production account. Do not expect Grafana to execute tests.