Selfhosted

Footprint numbers on other comparison pages had no reproducible method, so I read the releases and docs for all three on September 26, 2026 (Beszel v0.20.0, Netdata v2.11.1, Glances v4.5.7).

Only Netdata publishes one. Beszel's README says it's less resource-intensive than leading solutions and stops there. Glances says nothing. Netdata's resource-utilization page says 100-200 MB of RAM on an empty system and 250-350 MB in typical production, while its README says about 150 MiB and 5% CPU on production systems. Two pages, two answers.

History with nothing added differs a lot. Beszel keeps 1-minute points for an hour and averages up to 30 days. Netdata keeps per-second data up to 14 days, per-minute up to 3 months and per-hour up to 2 years, about 4 GiB of disk. Glances holds 1,200 values per stat in memory, about an hour.

Gotchas. Beszel v0.19.0 made agents verify the hub's HTTPS certificate, so a self-signed hub now needs CA_CERT_FILE. Issue #176 reports Beszel agents inside LXC showing the host's RAM. Netdata Cloud dashboards give access to 5 nodes without a paid plan (streaming to a Parent isn't capped), and Homelab was $90 a year when I read it.

I'd run a Beszel agent on every host and Uptime Kuma outside the network, with Netdata only on the box I'm debugging.

Every footprint figure here is the project's own claim, not my measurement. The page has a pidstat recipe for measuring your own: https://techfuelhq.com/homelab/beszel-vs-netdata-vs-glances-2026/