cloudparts

Netdata

A monitoring agent that collects thousands of metrics per second from each machine with no configuration, stores them locally, flags anomalies with per-metric machine learning and draws the dashboards for you.

What it is
Metrics
Used for
Alerting,Observability
Type
Open source
Made by
Netdata Inc.
License
GPL-3.0-or-later for the agent; the dashboard UI is closed source under the Netdata Cloud UI License (NCUL1)
Facts checked
23 October 2026
Website Source code

What it is

Netdata is an agent, written in Go, C and Rust, that you install on every Linux, Windows, macOS or FreeBSD machine. On start it discovers what is running, from CPU, disks, network, containers and systemd services to databases and web servers, through more than 800 integrations, and collects every metric once a second. Samples go into its own time-series engine on local disk, at about 0.5 bytes each, in three tiers: per-second, per-minute and per-hour, picked automatically by zoom level. For each metric it trains small k-means models on recent data and sets an “anomaly bit” on every sample that all models consider unusual. Hundreds of preconfigured alerts ship with it, and dashboards are generated from the metrics, so there is no query language to learn. The agent’s UI runs on port 19999.

Agents can stream to a Netdata “parent”, which centralises data for many children, keeps longer retention and runs their alerts and ML. Netdata Cloud, the company’s SaaS, does not store metrics: it queries the agents and parents live and adds multi-node dashboards, users, roles and alert routing. Metrics can also be exported to Prometheus (which can scrape /api/v1/allmetrics), InfluxDB, Graphite, OpenTSDB and others.

The licensing is split. The agent is GPL-3.0-or-later. The dashboard UI is a separate, closed-source package under the Netdata Cloud UI License, which allows its use only with Netdata’s own agents and Cloud and forbids modifying it. Netdata 2.0 (November 2024) moved the UI out of the repository into its own package, and 2.1 was announced as the last release with the old open-source dashboards; after it, only the v3 dashboard remains. The Cloud Community plan is free for personal use on up to 5 nodes; Business costs $4.50 per node per month.

Why teams pick it

  • Per-second metrics from the moment it is installed, with auto-discovery and generated dashboards instead of hand-built ones
  • Unsupervised anomaly detection on every metric, run at the edge without sending data anywhere
  • Data stays on your machines: tiered local storage, optional parents for central retention, and Netdata Cloud queries them in place
  • Light enough for small hosts (around 5% CPU and 150 MiB RAM by default, less with ML off) and exports to Prometheus or InfluxDB if you keep another stack

Watch out for

  • The UI is not open source: its licence allows use only with Netdata components, and distributions such as NixOS have disabled it as non-free
  • Free Netdata Cloud use is limited to 5 nodes for personal use; teams and commercial use need a paid plan, and viewing Windows nodes in the UI needs a paid subscription
  • Retention depends on each node’s disk unless you run parents, and a fleet-wide query fans out to every agent
  • Metrics-first: it reads systemd-journal and Windows event logs, but it is no replacement for a log store or distributed tracing

Alternatives

What all metrics have in common →Compare side by side →
TelegrafOpen sourceInfluxData’s plugin-driven Go agent that collects metrics from systems, services and devices and writes them to InfluxDB, Prometheus, Kafka and hundreds of other targets.

How it fits with other technologies

Let Prometheus scrape a Netdata agent (prometheus.yml) · yaml

scrape_configs:
  - job_name: 'netdata-scrape'
    metrics_path: '/api/v1/allmetrics'
    params:
      format: [prometheus]
    honor_labels: true
    static_configs:
      - targets: ['{your.netdata.ip}:19999']

See it in a real stack

No architecture uses Netdata yet. Drop it into the playground and wire it up yourself.Open playground →
Where it came from2013–2024

In 2013, at the company where Costa Tsaousis was COO, a large share of cloud transactions failed silently, and none of the monitoring tools they tried found the cause. “I couldn’t believe that monitoring systems provide so few metrics and with such low resolution, scale so badly, and cost so much to run,” he wrote, and he started building his own. The repository dates from June 2013; version 1.0, with per-second charts that updated live in the browser, came out in March 2016.

Tsaousis turned the project into a company, Netdata Inc., which raised a seed round around 2018 and builds Netdata Cloud on top of the agent. In September 2020 Bessemer Venture Partners led a $14.2 million round, bringing the total raised to $31 million, when the agent was being pulled from Docker Hub nearly a million times a day.

In July 2023 Netdata 1.41 bundled a new dashboard under the Netdata Cloud UI License, and Linux distributions objected: NixOS, for one, patched it out as non-free and kept the old open-source dashboard. Netdata 2.0 (November 2024) took the UI out of the repository as a separately licensed package, added native Windows support and announced the end of the old APIs and dashboards. By 2026 the agent had passed 80,000 GitHub stars.

Why it’s built this way

Per-second collection by default
Tsaousis’s complaint was low resolution: problems that last seconds vanish in one-minute averages. Netdata collects every second and keeps that tier on local disk.
Store and process at the edge
Keeping metrics on each node, or on a parent you run, avoids shipping and paying for all of them centrally. Netdata Cloud queries the agents rather than storing their data.
Dashboards and alerts that need no setup
Auto-discovery, generated dashboards and preconfigured alerts mean a single install is useful immediately, without writing queries.
A closed UI on an open agent
The company keeps the agent GPL but ships its newer UI as a binary licensed only for use with Netdata’s agents and Cloud, so the UI can’t be modified or pointed at other backends.

Timeline

  1. 2013

    Costa Tsaousis starts the project after an unsolved production failure

  2. 2016

    Netdata 1.0 released in March

  3. 2018

    Netdata Inc. raises a seed round

  4. 2020

    $14.2 million round led by Bessemer; $31 million raised in total

  5. 2023

    1.41 bundles the NCUL1-licensed dashboard; NixOS and others disable it

  6. 2024

    Netdata 2.0: native Windows, UI moved out of the repository

Sources12 · facts checked 23 October 2026

Improve this page on GitHub