See Every Container Heartbeat in Real Time

Netdata uses the cgroups.plugin to collect per-second metrics directly from Linux cgroups (v1 and v2). This provides CPU, memory, network, and disk I/O metrics for every container without requiring instrumentation. The docker collector (go.d) adds container state, health checks, and image metadata via the Docker API. Everything is auto-discovered - no configuration required.

Per-second monitoring captures the true behavior of your containers. Most operational issues last 2-10 seconds - CPU spikes, memory bursts, network congestion. Traditional 30-60 second monitoring averages these away, making them invisible. Netdata’s per-second granularity reveals transient problems that cause user-facing issues but disappear in averaged data. This is why we can reduce MTTR by 80%.

Yes, Netdata provides native Kubernetes support via Helm chart. Deploy with automatic DaemonSet (children) and Deployment (parents) architecture. Monitor nodes, pods, containers, services, deployments, and cluster resources with zero configuration. Label-based filtering, namespace isolation, and HPA tracking work out of the box. See Kubernetes monitoring for details.

Netdata charges predictable per-node pricing, not per container or metric. One agent monitors all containers on a host - unlimited containers, unlimited metrics, unlimited logs included. No data volume charges, no per-user fees. See pricing for P90 billing details and volume discounts.

For infrastructure monitoring, yes. Netdata provides per-second metrics (vs Prometheus 10-15s), automatic dashboards (vs manual Grafana building), and ML anomaly detection (vs manual alerting). No PromQL required - point-and-click analysis via NIDL framework. However, Netdata doesn’t yet support distributed tracing (planned Q2 2026). For APM workflows, pair with Jaeger or Tempo.

Netdata queries systemd-journal and Windows Event Logs directly - no log shipping, no indexing, no central clusters. This eliminates 90% of log infrastructure costs compared to ELK/Splunk. Logs correlate automatically with metrics for instant context. Analyze 1 million entries before sampling (200× more accurate than traditional tools). See zero-pipeline logs.

Netdata trains 18 unsupervised k-means models per metric on each agent. Models use different time windows and require unanimous agreement before flagging anomalies. This achieves 99% false positive reduction (theoretical rate: 10^-36). Anomaly Advisor then correlates findings across thousands of metrics to surface root causes in the top 30-50 results. No configuration, no cloud dependency, no data egress.

Netdata monitors Docker Swarm containers via the same cgroups mechanism as standalone Docker. However, there’s no dedicated Swarm orchestrator collector (unlike Kubernetes). You’ll get full container metrics but limited Swarm-specific metadata. For production orchestration, we recommend Kubernetes where Netdata provides native integration.

Netdata uses <5% CPU and 150-200 MB RAM per node (including ML models). The University of Amsterdam study (ICSOC 2023) confirmed Netdata as the most energy-efficient monitoring solution with the lowest overhead. For comparison, traditional monitoring can consume 10-30% CPU. Netdata’s efficiency comes from edge-based processing and optimized C/Go/Rust implementation.

Yes, Netdata Functions provide browser-based console replacement. Get real-time process monitoring (top), network connections (netstat), systemd services, and container logs - all with history and ML insights. Debug production containers safely without infrastructure access. This is unique to Netdata and eliminates the need for SSH during incidents.

Netdata automatically detects new containers within 1 second and removes stopped containers from dashboards. Metrics for stopped containers are retained according to your configured retention policy (default 14 days per-second, 3 months per-minute, 2 years per-hour). This ensures you can investigate issues even after containers are gone.

Netdata’s distributed architecture eliminates cardinality bottlenecks. Each agent stores its own metrics - no central aggregation, no memory explosions. Proven at 4.5+ billion metrics/second globally. Add containers freely without worrying about cost multiplication or performance degradation. This is fundamentally different from centralized systems like Prometheus.

Yes, Netdata provides full OTLP metrics and logs support via otel-plugin (Rust) and otelcol.plugin (Go). Distributed tracing (OTLP traces) is planned for Q2 2026. Netdata also natively scrapes Prometheus endpoints and exports via Remote Write. See integrations for details.

Yes, Netdata integrates with your existing stack. Export metrics to Prometheus, InfluxDB, Graphite, OpenTSDB, or TimescaleDB. Use the Grafana datasource plugin for existing dashboards. Send alerts to PagerDuty, Slack, Teams, Opsgenie, and 40+ other channels. Netdata complements or replaces tools based on your needs.

The open-source Netdata Agent is free forever with unlimited nodes, unlimited metrics, and community support. Netdata Cloud Community plan includes 5 nodes (personal, non-commercial use only) with multi-node dashboards, 1 user, and 4-hour alert retention. For production use, Business plan adds unlimited nodes, users, AI features, SSO, and enterprise support.