Netdata

What Netdata actually is

Netdata is an open-source, real-time infrastructure monitoring platform: an agent you install per node that collects metrics, stores them locally, runs anomaly detection on every series, and serves an auto-generated dashboard — plus Netdata Cloud, the commercial layer that ties nodes together in a browser. The project started in 2016 when Costa Tsaousis released it publicly; it picked up roughly 10,000 GitHub stars within a week of hitting Reddit, and the repository now shows over 80,000 stars with 690+ contributors. The agent is licensed GPLv3+, with the dashboard UI under a separate license.

The design is deliberately inverted from the usual observability topology. Instead of thin exporters shipping everything to a central time-series database, every Netdata Agent is a complete engine: collect, store, analyze, and detect anomalies at the edge, with Netdata Cloud coordinating access rather than ingesting your raw data. Netdata's own framing is data sovereignty — metrics stay on your infrastructure, and you access dashboards from anywhere. Whether that architecture suits you is the single biggest decision in adopting Netdata, so we will come back to it.

This is monitoring software, not a hosting product, so much of our usual spec grid (data centers, SLAs, per-plan cores) does not apply here. What we can assess is the pricing model, the maturity of the project, and where the free tier stops.

Company history matters for a tool you embed in your monitoring path. Netdata raised a $17 million Series A led by Bain Capital Ventures in September 2019, extended it with $14.2 million led by Bessemer Venture Partners in September 2020 for roughly $31 million total, and only introduced paid subscription tiers alongside the free community plan in February 2023. In other words, the commercial model is young relative to the project, while the free agent has been production software for a decade. We weight that combination positively: the parts you depend on are mature and open, and the parts you pay for are recent enough that pricing discipline still has to be proven over time. The team describes itself as remote-first and engineering-driven, around 30 people.

Pricing: what you really pay

Netdata's commercial pricing is per node, flat, and published openly — no ingest charges, no per-metric or per-seat metering, no retention tiers that double your bill when you keep data longer. The current plans on the pricing page:

• Community — Free. Up to 5 active connected nodes, secure remote access to your dashboards, centralized alert visualization, rooms organization, and customizable charts. Custom dashboards are limited to 1 per room and infrastructure-wide dashboards cap at 5 nodes. The data itself stays on your own systems.
• Business — $4.50 per node per month, billed annually (the page advertises a 25% saving for annual billing and shows an example total of $540.00/year at that rate). This adds unlimited metrics and logs, unlimited custom dashboards, unlimited data retention, Netdata AI for automated troubleshooting and reports, RBAC, SSO, SCIM, centralized configuration management, enterprise notification integrations (MS Teams, Slack, PagerDuty and others), and audit logs.
• Enterprise On-Premise — contact sales. Everything in Business with all components hosted on your premises, aimed at air-gapped facilities and critical infrastructure. It starts at 200 node licenses, so it is an enterprise conversation rather than an entry option.

Two things we like about this model. First, the free Community tier is not a time-limited trial: five nodes at per-second granularity with local storage is a real deployment, not a demo. Second, unit pricing is honest — a monitoring bill that scales with node count, not with how chatty your services are. Two things to watch: the interesting governance features (RBAC, SSO, SCIM) sit behind Business, so a security-conscious team with fewer than five nodes still has to pay to get role-based access; and the per-node cost multiplies quickly at scale, where Netdata's own enterprise pitch becomes the conversation. A Homelab tier is referenced by third-party listings between Community and Business, but it is not on the pricing page we reviewed, so we treat it as unconfirmed.

Performance & infrastructure

The agent is written mostly in C with components in Go, Rust, Python, and JavaScript, and runs on Linux, Windows, FreeBSD, and macOS. Collection coverage is broad: 800+ integrations by current counts, plus Prometheus endpoint scraping, OpenTelemetry ingestion over OTLP, StatsD, SNMP with vendor profiles and auto-discovery, an eBPF collector for kernel-level visibility, Azure Monitor auto-discovery, and a Nagios-compatible executor for existing plugin scripts. For log management, Netdata queries logs in place rather than shipping them into its own storage, which is where the "no ingest charges" claim comes from architecturally rather than as a discount.

Storage is Netdata's own dbengine time-series database with up to five storage tiers at different resolutions (per-second, per-minute, per-hour by default), claiming roughly 0.5 bytes per sample on disk at high resolution. The practical consequence: a node keeps days to weeks of high-resolution history locally without a separate storage bill, and long-term retention becomes an architecture question (Netdata Parents for aggregation, or external export) rather than a per-GB charge.

We have not run our benchmark suite against Netdata — our dated sysbench, fio, and iperf3 runs measure VPS plans (see /benchmarks), not monitoring agents — so we quote no agent overhead numbers here. What we can say from the architecture is that every collector runs locally per second, which is why Netdata can show sub-second-resolution incident detail that sampled systems cannot, at the cost of distributing storage and compute across every node you monitor.

Where it falls short

The distributed model has real operational edges. At scale you will want configuration management (Ansible, Terraform, or similar) to provision agents consistently — Netdata is not a set-and-forget install across thousands of hosts without tooling. Dashboard customization, while improved, remains more limited than the fully programmable panels of Grafana-style stacks, and the architecture assessment we referenced notes that monitoring a larger infrastructure effectively requires the commercial Netdata Cloud for centralized visibility.

The free tier's limits cluster exactly where growing teams feel them: one custom dashboard per room and a 5-node cap on infrastructure-wide dashboards. Above five active nodes you are on Business at $4.50 per node per month billed annually, and there is no published middle ground for a ten-person team that wants SSO without committing enterprise money — though annual billing at that rate is still modest against per-host observability pricing from the major SaaS vendors.

Support honesty is the weakest column in our rubric here. The Community tier is community-supported; the pricing page advertises priority customer support plans only on Enterprise On-Premise, and publishes no uptime or response SLA for Netdata Cloud. For a tool that sits in your incident path, that is a gap worth knowing about before you depend on the hosted layer.

Finally, there is lock-in of a mild but real kind: the agent is open source and the data is yours, which is genuinely good, but alerting, RBAC, and the collaborative workflow live in the commercial cloud product. Plan your exit path at the same time you plan your adoption.

Who should pick this

Our pick when you want maximum per-second visibility per euro and can tolerate operating the tooling. Homelabbers, small SaaS teams under five nodes, and DevOps groups that value data staying on their own machines get an unusually complete free product. The Business tier is fairly priced for teams that need SSO, RBAC, and AI-assisted troubleshooting without per-GB pricing anxiety — we would pick it over metered SaaS monitoring whenever node counts are stable and metric volume is high.

We would look elsewhere when you need a fully managed observability platform with published SLAs and enterprise support on every tier, when you want deeply programmable dashboards and a huge third-party plugin marketplace, or when your compliance regime requires a vendor-operated EU data residency you can point an auditor at. In those cases the mainstream SaaS observability vendors, or a self-hosted Prometheus/Grafana stack with contractual support, fit better.

Key takeaways

• Open-source (GPLv3+ agent), started in 2016 by Costa Tsaousis; 80,000+ GitHub stars and 690+ contributors today.
• Per-second metrics, local storage, and ML anomaly detection run in the agent itself — data stays on your infrastructure.
• Free Community tier covers up to 5 active nodes; Business is $4.50 per node per month billed annually with unlimited metrics, logs, and retention.
• Pricing is flat per node — no ingest, cardinality, seat, or retention charges, which keeps bills predictable.
• RBAC, SSO, and SCIM require the Business tier; Enterprise On-Premise starts at 200 node licenses for air-gapped environments.
• No published uptime SLA for Netdata Cloud, and priority support is an Enterprise item — factor that into incident-path dependence.

Sources

• https://www.netdata.cloud/pricing/
• https://www.netdata.cloud/about/
• https://en.wikipedia.org/wiki/Netdata
• https://dealroom.co/companies/netdata