​
Overview

Titled VM Instance Snapshots — “Point-in-time snapshots of the entire virtual machine including all volumes for backup, disaster recovery, and rollback”.
This section only exists if your account has the VM snapshot entitlement, which is off by default. Without it the section isn’t in the nav, and the Overview card reads Not enabled for this region.Ask support if you need it enabled. The feature works; its interface is rough, which is why it’s opt-in.
Navigation: Compute → Virtual Machines → click a machine → Snapshots.

​
Two sections, two kinds of snapshot

The page links across to the Volumes screen for per-volume work, because that’s where volume snapshots are normally created.

​
Taking one

Take Snapshot opens a dialog with a Snapshot Name and an optional Description, and a summary of what you’ll get: Full VM Snapshot (disk and memory) or Disk-only Snapshot (volumes only).
The button explains itself when it can’t be used, rather than just greying out:
A running machine cannot be captured disk-only. A disk-only capture of a live machine would be crash-consistent at best, so the platform requires memory to be included — or the machine to be stopped.

​
What holding snapshots prevents

While any VM snapshot exists on this machine, two operations are blocked:
  • Scaling the machine
  • Resizing any of its volumes
Both dialogs explain the block rather than failing obscurely. If you have resizing or scaling planned, delete the VM snapshots first.This is the main reason VM snapshots are a short-lived tool rather than a retention strategy — see Backups for something you leave in place.

​
Row actions

Transient chips Creating... and Deleting... appear while work is in flight.

​
Reverting

Reverting is destructive to current state, and the dialog says so:
  • The VM will be stopped during the revert process
  • All current data will be replaced with the snapshot state
  • Any changes made after this snapshot will be lost
  • The VM will need to be restarted after reverting
There is no undo. If the current state might matter at all, take a fresh snapshot before reverting to an old one — then you can revert the revert.

​
Common task: a safe in-place upgrade

1

Snapshot with memory, while running

Take Snapshot → name it for its purpose, e.g. pre-upgrade-2026-09. Include memory so you can resume exactly.
2

Wait for it to complete

A snapshot still Creating is not a restore point. Don’t start the upgrade until it’s ready.
3

Do the upgrade

Apply your changes.
4

If it went wrong, revert

Revert to Snapshot. The machine stops, returns to the captured state, and needs starting again.
5

Delete the snapshot once you're satisfied

This step is not optional housekeeping. While it exists it consumes storage quota and blocks scaling and volume resizes.
For rolling back a single disk, a volume snapshot is the lighter choice — no entitlement needed, and it doesn’t block scaling. Reach for a VM snapshot when you need memory state, or all of a machine’s disks captured as one consistent set.

​
Empty and error states

VM Snapshots

The account-wide screen and the entitlement.

Backups tab

Protection you leave in place.

Volumes tab

Why resizing is blocked while snapshots exist.