> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rafftechnologies.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Storage model
> How base disks, volumes, snapshots, backups, and custom images relate — and what survives what
Updated May 8, 2026
A Raff VM has up to **five storage objects** in its orbit, and the source of most "where did my data go?" support tickets is mixing them up. This page is the map.
## The three categories
| Category | Purpose | Lives where |
| ---------------------- | ----------------------------------------- | -------------------------------------------- |
| **Active storage** | The disks the running OS reads and writes | Attached to a single VM at a time |
| **Protective copies** | Point-in-time captures for restore | Stored in the platform, separate from the VM |
| **Reusable templates** | Configurations you can boot new VMs from | Stored at the account level |
Every storage object on a Raff VM falls into exactly one of these.
## Active storage
These are what the OS sees as block devices.
### Base disk
The boot disk that comes with every VM. Sized by the plan you pick (e.g. 50 GB on the \$9.99/mo plan). Holds the OS, installed software, and anything you didn't put on a volume.
* Size grows via [resize](/products/build/virtual-machines/quickstart-guides/resize) → Storage tab. **One-way** — disks can grow but cannot shrink.
* The base disk and the VM are inseparable: deleting the VM destroys the base disk.
### Block volumes
Additional storage you attach on top of the base disk — see [Volumes](/products/store/volumes).
* Live in the same region as the VM; can be detached and reattached to a different VM in that region.
* Survive the VM. Delete the VM → the volume detaches but is **not** deleted; it keeps billing until you delete it explicitly.
* Format and mount inside the OS — see [Attach to a VM](/products/store/volumes/quickstart-guides/attach-to-vm).
## Protective copies
Captures of disk state for restore. Different lifecycle, different retention, different price.
### Snapshots
Manual, on-demand point-in-time copies of a VM disk or a volume.
* Created when you click **Snapshot** — see [Create a snapshot](/products/build/virtual-machines/quickstart-guides/create-snapshot).
* Stored until you delete them, or until the source VM is destroyed (VM-disk snapshots are deleted with the VM; volume snapshots survive).
* Billed per **GB-hour** — accrues continuously, settled at month-end on the 1st.
* Best for restore points before risky changes.
### Backups
Scheduled or on-demand copies, governed by a retention policy.
* Configured per-VM with a Daily / Weekly schedule and a retention window (1-365 days). See [Enable backups](/products/build/virtual-machines/quickstart-guides/enable-backups).
* Backup + snapshot storage shares a **free pool per account** (50 GB + 10–15% of each VM's disk).
* Above the pool: \$0.05/GB/month on the combined excess only.
* Crucially: **backups survive VM delete** as long as the retention is still active. This is the only mechanism for true post-delete recovery.
For when to use which, see [Snapshots vs Backups](/products/build/virtual-machines/concepts/snapshots-vs-backups).
## Reusable templates
Configurations from which you can deploy *new* VMs.
### Stock OS templates
Raff-maintained images: Ubuntu, Debian, Fedora, CentOS, AlmaLinux, Rocky, openSUSE, Alpine, FreeBSD, Windows Server. See [OS templates](/products/build/virtual-machines/concepts/os-templates).
### Custom images
A snapshot of a VM's OS disk preserved as a reusable template you can launch new VMs from. See [Custom images](/products/build/virtual-machines/concepts/custom-images).
* Outlive the source VM (unlike snapshots).
* Appear in the create-VM dialog alongside stock templates.
* Best for golden-image workflows: build once, deploy N times.
## Survival matrix — the key reference
What survives each lifecycle event:
| | Base disk | Attached volumes | Volume snapshots | VM snapshots | Backups | Custom images |
| ------------------------- | ----------- | ---------------- | ---------------- | ------------ | -------------------- | ------------- |
| **Stop / start** | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| **Reboot** | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| **Resize compute** | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| **Resize disk (grow)** | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| **Reset password** | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| **Reinstall (Change OS)** | ❌ wiped | ✅ | ✅ | ❌ deleted | ❌ deleted | ✅ |
| **Factory reset** | ❌ wiped | ✅ | ✅ | ❌ deleted | ✅ | ✅ |
| **Delete VM** | ❌ destroyed | ✅ detached, kept | ✅ | ❌ deleted | ✅ kept per retention | ✅ |
Reading the matrix:
* **Backups are the only thing that survives delete-VM.** If you might want to recover the OS disk after destroying the VM, you must have a backup — not a snapshot.
* **Reinstall is more destructive than factory reset.** Reinstall deletes both snapshots *and* backups; factory reset preserves backups.
* **Volumes always survive** any VM-level operation; only volume-level delete removes them.
## Pricing summary
| Object | How it's billed |
| -------------------------------------------------- | ------------------------------------------- |
| Base disk | Bundled in the plan price |
| Block volumes | Per GB stored — subscription term or hourly |
| Snapshots | Per GB-hour |
| Backups | Per GB-month above the shared free pool |
| Custom images | Per GB stored |
| Object Storage (orthogonal — for files, not disks) | Per GB stored |
All charges accrue continuously, settle on the 1st of the month, and deduct from your account balance first before charging the saved card. See [Pricing](/products/build/virtual-machines/details/pricing) for the full picture.
## "I need to…" — decision flowchart
| Goal | Use |
| -------------------------------------- | ----------------------------------------------------------------------------------------------------------------- |
| Add more disk to one VM | Resize the base disk **or** attach a [volume](/products/store/volumes) |
| Capture state before a risky change | [Snapshot](/products/build/virtual-machines/quickstart-guides/create-snapshot) |
| Routine, scheduled protection | [Backups](/products/build/virtual-machines/quickstart-guides/enable-backups) |
| Move a VM to a different region | Snapshot → Save as image → create new VM in target region from image |
| Stand up N identical VMs from a config | Build one → save as [custom image](/products/build/virtual-machines/concepts/custom-images) → create N from image |
| Recover a destroyed VM's OS disk | Only possible from a backup — contact support |
| Keep data when the VM is destroyed | Put it on a [volume](/products/store/volumes) (volumes survive); the OS disk doesn't |
## Related
Decision framework for protective copies.
Save and reuse a VM's OS disk.
Block storage that outlives the VM.
Per-storage-object cost details.