How to back up a cloud dev machine, and what persists on its own

A persistent machine keeps your work between sessions. That is not the same as a backup. Persistence protects you from closing a tab; a backup protects you from a mistake. You want both, and they are different habits.

What persists on its own

Your files, your installed packages, your configuration, and your running processes all stay put across disconnects and reboots. You do not have to do anything to get that; it is how the machine works.

What a backup adds

• A copy you can restore after you delete the wrong thing.
• A point in time you can compare against or roll back to.
• Somewhere off the machine your work also lives.

Practical habits

The best backup for code is the one you already use: push to a remote often. For everything that is not in git — data, generated artifacts, configuration you tuned by hand — an rsync to storage you control covers the gap. That one is a habit rather than a feature: decide the cadence once and let it run.

Some plans are also built with the provider's own automated backup — the pricing table says which. It is a snapshot held in the provider's systems rather than a copy you hold, there is no restore you can start yourself, and nothing has been committed about how far back it reaches. Useful, and not a reason to skip the habit above.