Sees every metric every second, then uses per-metric ML to flag what looks abnormal
GPL-3.0
- C
- Go
- Rust

About Netdata
Netdata trades sampling intervals for resolution: it charts system and application metrics at one-second granularity, so a spike that a one-minute tool would average away shows up immediately. Install it and dashboards populate themselves with little configuration, which is how it stays useful for lean teams as much as large fleets.
It collects from systems, containers, apps, logs, APIs, and synthetic checks into a tiered time-series database. A model trained per metric learns recent behavior and surfaces anomalies, alert rules drive notifications, agents stream to Netdata Parents, and metrics export to Prometheus, InfluxDB, OpenTSDB, and Graphite. The agent UI lives at localhost:19999.
Netdata runs on Linux, macOS, Windows, FreeBSD, Docker, and Kubernetes under GPL v3. Netdata Cloud is optional, and data can stay on your own infrastructure with no central collection.
Key features
- Per-second metrics and visualizations
- Collects from systems, containers, apps, logs, APIs, and synthetic checks
- ML models per metric for anomaly detection
- Built-in alerts with email, Slack, Telegram, PagerDuty, Discord, and Teams
- Streams to Netdata Parents and exports to Prometheus, InfluxDB, OpenTSDB, and Graphite
Details
- First released
- 2013
- Platforms
- Windows · macOS · Linux · Docker
- Deployment
- self-hostable · docker · cloud
- Storage
- Tiered time-series database
- Governance
- CNCF · Netdata Inc
- License
- GPL v3
