Idle RAM by self-hosted category: no-code databases lead at about 1.4 GB, file sharing sits near 32 MB

copySelfHost Atlas (2026). Idle RAM by self-hosted category: no-code databases lead at about 1.4 GB, file sharing sits near 32 MB. https://selfhostatlas.com/blog/idle-ram-by-category. Data: doi:10.5281/zenodo.22983606

If you are choosing what to self-host, it is tempting to ask which kind of software is heavy. We grouped our idle-RAM measurements by category to see how far that question gets you. The medians do differ. The ranges inside most categories are wider than the gaps between medians.

#Median idle RAM by category

These are the 22 categories with at least four measured apps. Together they cover 118 of the 194 apps with a measured idle RAM; the other 41 categories have fewer than four measured apps and are left out because a median of two or three apps says very little. The count below is measured apps in that category, which can be lower than the number of apps we list there.

For reference, the median across all 194 measured apps is 174.5 MB. Twelve of the 22 categories sit above it.

#The spread inside a category is the bigger story

In 19 of the 22 categories, the heaviest app idles at ten times or more the lightest one in the same category. A few examples from the table:

• Web analytics: 138 MB to 16,824 MB. The heaviest is PostHog.
• Project management: 26 MB to 3,699 MB. The heaviest is Huly.
• Team chat: 110 MB to 3,259 MB. The heaviest is Zulip.

So the category does not pin down a RAM figure for most of them. The specific app does, and the per-app figure is on each app's page.

#Read the medians with care

With four to eleven apps per category, a median moves when one app is added or re-measured, and in a category of four it is the average of the middle two. Treat the ordering as a rough picture, not a ranking. The medians at the top and bottom are far apart (no-code databases and file sharing); the middle of the table is crowded, and the ranges overlap.

#What these numbers are

Idle RAM is the memory docker stats reports for every running container of the app, summed, 30 seconds after the install finishes, on a server pruned of earlier apps. It is a 30-second snapshot, not steady-state usage, not a peak, not under load and not a sizing recommendation. 188 of the 194 measurements were taken on a Google Cloud e2-standard-2 running Ubuntu 26.04 LTS with Docker; six apps (PostHog, Sentry, Langfuse, RustDesk, AppFlowy and LiteLLM) were run on larger e2-standard machines. How the harness works is described on How we test. The per-app extremes across the whole set are in the heaviest-and-lightest post.

#About the numbers

This post uses the install-verification dataset exported on 2026-10-15 (256 records, 194 with a measured RAM figure, measured between 2026-08-28 and 2026-10-14). Categories are the dataset's own category field. The dataset is CC BY 4.0 and cited as DOI 10.5281/zenodo.22983606.