Why I stopped trusting the cloud with my own data

It started with a price increase email. Not a big one — a couple of dollars a month for storage I was barely using. But it made me open the folder and actually look at what was in there: every photo since 2018, tax documents, the backups of three client databases. All of it sitting on a machine I had never seen, governed by terms I had agreed to without reading.

What I actually moved

Not everything, and that matters. Email stays where it is — running your own mail server is a hobby that eats other hobbies. What came home was the stuff where availability matters less than ownership: file backups, notes, media, and the client database dumps that legally should not live on someone else's hardware anyway.

The test I settled on: if losing access for a day would ruin my week, it stays hosted. If losing control of it forever would ruin my year, it comes home.

The part nobody mentions

You become the on-call engineer. A drive fails on a Sunday and there is no support ticket to open — there is only you, a screwdriver, and the backup you hopefully tested. I test mine now, monthly, because the first time I needed a restore I discovered the job had been silently failing for five weeks.

# the cron job that actually saved me
0 3 * * * restic backup /srv --tag nightly \
  && restic check --read-data-subset=5%

That second line is the whole lesson. Taking a backup is easy and feels productive. Verifying it is boring and is the only part that counts.

Was it worth it

Yes, but not for the reasons I expected. It is not cheaper once you count my time. It is not more reliable than a datacenter. What it is: mine. When something breaks I understand why, and fixing it teaches me something I keep. That is the same reason I repair my own washing machine, and it is probably not a rational argument.