Workspace Backups & Migration
Snapshot a directory on a running GPU box, then restore it onto a different GPU (or a different provider) and pick up where you left off.
Snapshots, automated or on demand
A workspace backup is a snapshot of a directory on a running GPU box that you can later restore onto a different box: a different GPU, a different provider, a different region. Nothing is scheduled for you automatically at deploy time: automated snapshots are opt-in, and until you turn them on for a deployment, the only backup that exists is one you take by hand.
Turn it on and point a deployment at an absolute path and an interval: the console form enforces a 15-minute minimum with a 30-minute default; going through the API or the CLI instead, the minimum is 1 minute. You can also take a snapshot by hand at any moment, before shutting a box down, for instance, which is the one habit that makes everything else here work regardless of whether automated snapshots are on.
Snapshots are incremental: the first one carries the directory, later ones carry what changed. A snapshot taken seconds after another with no writes in between is effectively free, in both time and stored bytes.
Every snapshot lands in a backup history for that deployment with its path, timestamp, size, and whether it was automated or manual.
A restore is not tied to the box it came from
This is the part that matters. A snapshot is not welded to the instance that produced it. When you deploy a new box, you can choose an existing snapshot to restore into it: a different GPU model, a different region, a different provider entirely.
So "migrating" a workload is not a special feature with its own machinery. It is three steps you already know: snapshot the box you are on, launch a box where you want to be, restore the snapshot onto it.
Why this is the point of a multi-provider marketplace
Comparing prices across providers only helps if you can act on what you find. Without portable state, a cheaper GPU somewhere else is a rebuild: reinstall CUDA, re-pull weights, re-create the environment, lose a morning. That friction is what keeps people on whichever cloud they set up first.
Portable snapshots are what turn a price difference into a decision you can make on a Tuesday afternoon.
Where the data goes
Aquanode's own storage is the zero-configuration default and needs no credentials from you. Backups and restores work out of the box. Taking a snapshot is free; holding one in Aquanode Storage is billed at $0.02 per GiB per 30-day month, metered per second. Writing to a bucket you own instead (your account, your region, your keys) carries no Aquanode storage charge and is part of the enterprise plan; book a demo and we set it up with you.
Backup and migration fees are listed in the console next to the storage settings, separately from GPU cost, so a restore never shows up as a mystery line on a GPU bill.
FAQ
Can I restore a snapshot onto a different GPU?
Yes. A restore is not tied to the deployment that created the snapshot: when you launch a new box you can select an existing snapshot to restore into it, on a different GPU model, region, or provider.
Are snapshots full copies every time?
No. Snapshots are incremental: the first carries the directory and later ones carry only what changed, so a frequent schedule is cheap in both time and stored bytes.
Can backups run on a schedule?
Yes, but only once you turn it on: nothing is scheduled automatically at deploy time. Give a deployment an absolute path and an interval and Aquanode snapshots it on that schedule; the console form's minimum is 15 minutes with a 30-minute default, and the API or CLI accepts as low as 1 minute. You can also take a manual snapshot at any point, and both appear in the same backup history.
If a provider kills my box, can I recover right up to that moment?
Only as good as your last snapshot. If you stopped the box yourself, it was captured on the way out and you can restore it anywhere: that's always claimable. If the provider reclaims a running box out from under you (a spot reclaim, a node failure) you can only recover back to a snapshot that was already scheduled or already taken by hand, never to the instant it died. Turn on automated snapshots if you want that window to be short.
Is snapshot storage billed separately?
Taking a snapshot is free. Holding one in Aquanode Storage is billed at $0.02 per GiB per 30-day month, metered per second: that is the zero-config default, and it covers backups, restores and cross-provider migration. Pointing backups at a bucket you own (your S3, Backblaze B2 or MinIO account, your region, your keys) carries no Aquanode storage charge and is part of the enterprise plan; book a demo and we set it up with you.
How do I move a running workload to a cheaper provider?
Snapshot the directory that holds your environment, launch a new instance on the provider you want, and restore that snapshot onto it. There is no separate migration product: the portable snapshot is the mechanism.
