Grafana
Dashboards and alerts on top of your metrics, logs and databases
Alternative toDatadog
Grafana turns data from sources such as Prometheus, Loki, PostgreSQL or InfluxDB into dashboards, and sends alerts when a value crosses a line. The open source edition is AGPL-3.0 and self hostable, and Grafana Labs sells a hosted cloud and enterprise features.
Best for when
- you want graphs and alerts for your servers and containers
- you already run Prometheus, Loki or a SQL database worth visualising
- you want one dashboard tool that talks to many data sources
Watch out for
- it is only the display layer, so you still need to run and store the data behind it
- each data source has its own query language to learn
- some features are reserved for the paid enterprise and cloud editions
One screen for everything you measure
Grafana does not collect data itself. It connects to data sources and lets you build panels: CPU over time, request latency, disk space, rows in a table, anything a query can return. Ready made dashboards for common exporters can be imported, which gets a home server overview running in minutes once the metrics exist.
It suits anyone who already has, or plans to set up, a metrics store such as Prometheus and wants graphs and alerts in one place. The usual home stack is Prometheus for metrics, node exporter and cAdvisor for host and container stats, and Grafana in front.
What running it involves
Grafana alone is moderate: a few hundred megabytes of RAM and a small SQLite database for dashboards and users. The real footprint is the stack behind it, because metrics and logs storage grows with retention and with the number of things you measure.
Keep it behind a VPN or a login you trust, change the default admin password on first start, and back up the database or provision dashboards as files. Building good dashboards is a skill of its own, and each data source has its own query language.
