See Every Node in Your Fleet, Every Second, From the Edge

A lightweight Netdata Agent runs on each endpoint, collecting per-second metrics, training ML models locally, and storing data in a built-in time-series database. Agents stream to regional Netdata Parents for aggregation and longer retention. Netdata Cloud provides unified dashboards and cross-fleet alerting — receiving only metadata, never bulk telemetry. New to the model? Read what is edge monitoring, or learn how the Agent works.

Each Agent continues collecting, analyzing, and storing data locally during outages. When connectivity returns, the Agent backfills the Parent with all buffered data — no gaps, no blind spots. Clustered Parents provide additional high availability. See how Parents work.

Agents use a multi-tier time-series database storing per-second data at approximately 0.6 bytes per sample. Default retention covers 14 days of per-second data, 3 months of per-minute, and 2 years of per-hour aggregations — all in gigabytes, not terabytes.

A standalone Agent uses under 5% of a single CPU core and 150-200MB RAM, including ML training. In child mode (streaming to a Parent with ML and alerts offloaded), usage drops to under 2% CPU and 100-150MB RAM with zero disk I/O.

Netdata uses simple per-node pricing. Agents are free and open source. Cloud plans bill per active node with P90 billing protection — brief daily spikes and your top 3 days each month are excluded. Costs scale linearly with your fleet, not with data volume. See pricing.

Yes. Netdata includes 800+ auto-discovered collectors covering SNMP devices, Modbus sensors, 1-wire devices, GPUs, IPMI hardware sensors, and custom metrics via StatsD, OpenMetrics, and OpenTelemetry. For IoT gateways and edge devices specifically, see IoT and edge device monitoring. Explore integrations.

No. Netdata generates dashboards algorithmically based on the data each node exposes. Every chart supports point-and-click slicing by node, instance, dimension, and label — no query languages like PromQL required. Explore algorithmic dashboards.

Netdata Functions surface process lists, network connections, systemd units, disk stats, and IPMI sensors directly in the browser — with per-second precision and historical context. Systemd-journal logs are searchable without a separate log pipeline. Explore troubleshooting features.

Yes. Netdata Cloud is available as an on-premises deployment for air-gapped environments. Agents and Parents operate independently with local dashboards even without internet connectivity. Explore on-premises options.

Each Agent trains 18 unsupervised ML models per metric locally, requiring zero configuration. Models retrain every 3 hours and achieve 99% false positive reduction through consensus. The Anomaly Advisor correlates anomalies across all metrics on a node to surface root causes. Explore ML features.

Netdata supports one-line kickstart installation, native packages for all major Linux distributions, Docker containers, Kubernetes Helm charts, and Windows MSI installers. Configuration is text-based and version-control friendly for Ansible, Terraform, and other provisioning tools.

Parents aggregate streams from multiple Agents, provide longer retention and high availability, and sit close to the fleet to minimize latency. Netdata Cloud unifies dashboards, RBAC, alert routing, and distributed queries across all Parents and Agents — receiving only metadata, never metric data. Learn about the architecture.

Yes. Netdata Cloud supports multi-tenant isolation through Spaces (physical separation per customer) and War Rooms (logical separation per team or service). Each Space has its own users, billing, and infrastructure. Parents can be dedicated per customer for complete data isolation. Explore MSP solutions.