> This page is for Data Protection.

> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.eon.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.eon.io/_mcp/server.

# Compute Engine: Restore a VM (Instant Restore)

Use instant restore when you need the VM back in service before all its disk data is restored.
Instant restore makes the instance available while disks hydrate in the background, serving data on demand until all data is restored.

This article explains how to restore a Google Compute Engine virtual machine (VM) from your Eon snapshots using instant restore.
For the standard restore method that fully restores every disk before the instance becomes available, see [Compute Engine: Restore a VM].

[Compute Engine: Restore a VM]: /data-protection/restoring/restore-cloud-workloads/gcp/compute-engine/compute-engine-restore-a-vm

## Restore the instance with instant restore

#### Before you start

Before you restore a Compute Engine VM with instant restore, make sure:

* A [restore account has been connected] to your Eon account.

* The GCP restore project has a VPC and subnet that meet [restore server connectivity requirements].
  By default, Eon uses the default VPC and subnet, but you can [customize the restore server connectivity] if needed.

[restore account has been connected]: /data-protection/restoring/restore-accounts/about-restore-accounts

[restore server connectivity requirements]: /data-protection/restoring/restore-accounts/about-restore-accounts#the-restore-server-and-its-network-connectivity

[customize the restore server connectivity]: /data-protection/cloud-workloads/gcp/restore-projects/gcp-customize-restore-server-configuration

* Your restore project has restore permissions version 1.7.3 or later.
  See [GCP: Permissions Deployment Files] for deployment files.

* Your snapshot and source VM meet the requirements in [Restore support notes].

* The subnet you plan to use for the restored VM is different from the temporary restore server's subnet.
  The restored VM runs as a nested VM inside the restore server, which has a separate network interface in each subnet.
  See [GCP: Customize Restore Server Configuration] to choose a separate subnet for the restore server.
  For Google Cloud's network requirements, see [Create VMs with multiple network interfaces].

[GCP: Permissions Deployment Files]: /data-protection/cloud-workloads/gcp/permissions/gcp-permissions-deployment-files

[Restore support notes]: /data-protection/cloud-workloads/gcp/resources/compute-engine#restore-support-notes

[GCP: Customize Restore Server Configuration]: /data-protection/cloud-workloads/gcp/restore-projects/gcp-customize-restore-server-configuration

[Create VMs with multiple network interfaces]: https://cloud.google.com/compute/docs/instances/create-instance-multiple-nics

#### Find the snapshot you want to restore

1. In your

   inventory

   , select the Compute Engine resource you want to restore.
   The **Resource details** panel opens.

2. Click the **Eon snapshots** tab.
   You'll see a calendar view with colored dots on the days when snapshots were taken.
   There can be more than one snapshot per day.

   When you click a day, a list of Eon snapshots taken that day is displayed below the calendar.

3. Expand the snapshot you want to restore.
   Click **Instance**, and then click **Restore**.
   The **Restore GCE instance** wizard opens.

#### Choose the restore account

In the **Restore GCE instance** wizard, you're taken through a three-step process.
The first step is to choose a restore account.

Choose the cloud account you want to restore to, and then click **Next**.

#### Turn on instant restore and set the instance details

[Machine families resource and comparison guide]: https\://cloud.google.com/compute/docs/machine-resource

In the **Instance configuration** step, you'll set the instance details.

1. Turn **Use instant restore** on.

   With instant restore, the instance restores to the same region as your vault, and **Region** selection isn't available.
   The instance starts automatically, so **Start instance after restore** isn't displayed.

2. Set the **Machine name**.

3. Choose a **Zone** to deploy the restored instance to.

4. Choose the **Machine type**.
   With instant restore, the list includes only machine types in the C2, C3, C4, N2, N4, and N4D families.
   For information on available options, see [Machine families resource and comparison guide] from Google Cloud.

5. Choose a VPC **Network** and **Subnetwork** for the restored VM.
   This subnet provides the nested VM's network identity, so it must be different from the restore server's subnet.

6. If you want to retain the labels that were present when the snapshot was taken, turn **Keep original labels** on.
   Otherwise, turn **Keep original labels** off.

   In both cases, the new machine is labeled with `eon-restore=true`.
   Labels with the `eon-` prefix aren't included in the original labels because Eon uses this prefix for internal operations.

7. To specify an internal IP address, turn **Assign internal IP automatically** off and enter an **Internal IP address** within the selected subnet.
   Otherwise, leave the switch on to assign an available address automatically.

8. Click **Next**.

#### Set details for the attached disks

In this step of the wizard, you'll set encryption, disk settings, and labels for the disks attached to the instance.
Expand individual disks to show disk-specific settings.

#### Encryption

You can encrypt all disks with the same customer-managed encryption key (CMEK), or you can set each disk's key separate from the other disks.
If you choose to encrypt each disk separately, turn off **Encrypt all disks with the same key**.
Encryption settings are then displayed for each disk.

#### Disk settings

For each disk, you can keep the original disk settings, or you can customize the drive's name, type, and size, and, for disk types that support them, IOPS and throughput.

#### Labels

Each disk has its own labels, independent of the instance-level labels in the previous step.
If you want to retain the labels that were present when the snapshot was taken, turn **Keep original labels** on.
Otherwise, turn **Keep original labels** off.

In both cases, the restored disk is labeled with `eon-restore=true`.
Labels with the `eon-` prefix aren't included in the original labels because Eon uses this prefix for internal operations.

#### Start the restore job

Click **Restore** to start the asynchronous restore job.

You can monitor the restore job on the

Restore jobs tab

in the Jobs page.

The job shows **Complete** when the instance becomes available, while disks continue hydrating in the background.

Once hydration completes, the instance continues running normally.
When you reboot or shut down the instance, it's promoted to the machine type and configuration you specified during restore.
The instance keeps its name and internal IP address through this transition.
If you shut down the instance, it stays stopped after promotion.
If you don't reboot or shut down the instance within 14 days after hydration completes, it isn't promoted and keeps running as is.