Limitations

This section outlines important policies and infrastructure limitations for Cloud GPU VMs.

Deployment & availability

• Frankfurt Data Center: You can currently access Cloud GPU VMs only in the Frankfurt de/fra/2 data center.

Instance lifecycle policy

• Maintenance reboots: Cloud GPU VMs might restart without prior notification to apply system updates, security patches, or infrastructure maintenance. Design your applications to handle unexpected restarts and configure services to restart automatically on boot.

• No Live Migration: Cloud GPU does not support live migration for VMs. Maintenance events might stop or reboot VMs.

• Fixed resource scaling: Size your servers appropriately at creation time. CPU core and RAM upscaling or downscaling are not currently supported after provisioning.

Management restrictions

• API only support: You can manage Cloud GPU VMs only via the Cloud API. You cannot use the Data Center Designer (DCD) or other interfaces for management.

• Templates: You cannot create new templates or modify existing templates.

• Images: The platform supports IONOS Linux images only at launch.

• Boot volume restrictions: The first connected boot volume has the following restrictions:

  • It cannot be detached from the VM.

  • It cannot be upscaled after deployment.

  • Size must be determined during the initial provisioning.

• Resource scaling restrictions: Size servers appropriately at creation time, because CPU core and RAM upscaling or downscaling are not currently supported. Plan your workload requirements carefully during the initial provisioning.

• Volume Management: Additional volumes support standard management operations, including scaling up, detaching, and attaching across the Cloud API interface.