MEASUREMENTS

Everything published here, and how to check it

Each of these answers one question a home-server owner has to answer anyway, by running the thing rather than by summarising other pages about it. Each one ships its script, its machine, and its raw output.

Published

• What backup compression actually buys a home server
  Thirteen settings across five kinds of data on a 282 MB corpus you can rebuild. The slowest setting spends 360 times longer than the fast one to save 6.8% more, because half the bytes on a home server are photographs and video that compress by nothing at all. Includes the restore-speed column that ratio comparisons leave out. 21 August 2026.
• What SQLite's durability settings cost on a home server
  Thirty-six configurations of journal_mode, synchronous and fullfsync, measured as insert throughput. On macOS, asking the drive to genuinely flush costs a factor of ninety — and the default is not to ask. 21 August 2026.

How to check any of it

The raw data page holds every script and every line of output behind these pages, along with the specification of the machine they ran on. Nothing here is a figure you have to take on trust: rerun the script, and if your numbers disagree, that result is more useful than the one published and it belongs in the inbox.

What is not here

No roundups, no "best self-hosted app" lists, no news. Publishing happens when a measurement is finished, which means there are quiet months. That is the intended behaviour rather than a lapse — the about page explains why, along with who writes this and how it is funded.