I spent a weekend setting up Grafana + Prometheus + node_exporter + cAdvisor to monitor 4 servers. It worked, but the Prometheus instance alone ate 300MB of RAM, and every time I wanted a new dashboard I had to write PromQL queries.
Then I found Beszel.
What Is Beszel?
Beszel is a lightweight, open-source server monitoring tool built as a single Go binary. It tracks CPU, memory, disk, network, and Docker container stats through a clean web dashboard.
The agent uses under 10MB of RAM per server. Compare that to Prometheus at 300MB+ or Netdata at 200-500MB. On my Raspberry Pi 4, that difference is the gap between “monitoring works” and “monitoring kills the workload.”
Setup: 5 Minutes, Zero Config
The setup is genuinely fast:
- Run the Beszel hub (single Docker container)
- Copy the agent install command from the web UI
- Run it on each server you want to monitor
No YAML files. No scrape configs. No dashboard JSON imports. The dashboard is pre-built and shows exactly what you need: CPU usage, memory, disk, network throughput, and per-container Docker stats.
What Beszel Monitors
| Metric | Details |
|---|---|
| CPU | Total and per-core usage with historical graphs |
| Memory | Used, cached, available with trend lines |
| Disk | Usage per mount point, I/O stats |
| Network | Bandwidth in/out per interface |
| Docker | Per-container CPU, memory, and network usage |
| Temperature | CPU and GPU temperatures |
| Alerts | Threshold-based notifications for any metric |
Historical data goes back days and weeks, so you can spot trends like “memory usage has been climbing 2% per day” before things crash.
Beszel vs the Alternatives
I’ve tried most of the popular monitoring tools. Here’s how they compare:
Beszel vs Grafana + Prometheus: Grafana is infinitely customizable, but that flexibility comes with complexity. Beszel gives you 90% of what most people actually look at (CPU, RAM, disk, Docker) with zero configuration. If you need PromQL queries and custom data sources, stick with Grafana. If you just want to know if your servers are healthy, Beszel is the simpler choice.
Beszel vs Netdata: Netdata offers 2,000+ metrics at 1-second resolution with AI anomaly detection. It’s impressive but uses 200-500MB RAM per node and overwhelms you with charts. Beszel trades depth for clarity – fewer metrics, but the ones you actually check daily.
Beszel vs Dozzle: These solve different problems. Beszel monitors server metrics; Dozzle shows Docker container logs. Most self-hosters run both together – Beszel tells you something is wrong, Dozzle helps you read the logs to find out why.
The Monitoring Stack I Actually Use
After trying everything, my monitoring setup is:
- Beszel for server and Docker metrics (CPU, memory, disk, alerts)
- Dozzle for Docker container logs
- Uptime Kuma for external endpoint monitoring
Total RAM usage across all three: under 50MB. Setup time: about 15 minutes for everything. Compare that to my old Grafana stack at 800MB+ and a weekend of configuration.
Who Should Use Beszel?
Beszel works best for:
- Homelab users monitoring a handful of servers and Docker containers
- Small teams that want server monitoring without a dedicated DevOps person
- Developers running side projects who need alerts when things go wrong
- Anyone switching from Netdata who found it too heavy or overwhelming
If you’re running enterprise infrastructure with 100+ nodes, custom metric pipelines, and a full SRE team – use Grafana, Datadog, or Netdata. They’re built for that scale.
For everyone else, Beszel is monitoring that just works.
Getting Started
Beszel is free and open source (MIT license). You can self-host it on your own server or deploy it on InstaPods for $3/mo if you don’t want to manage the infrastructure.
The GitHub repo is at github.com/henrygd/beszel.
