Today, we're excited to announce that hibernation for Azure GPU VMs is now generally available. We announced the extension of hibernation support to GPU Virtual Machines (VMs) in Azure in April 2024. This feature allows for significant savings on compute costs during idle times, enabling users to pause their GPU-powered VMs while preserving the machine's in-memory state. Customers incur only storage and networking charges, and when needed, they can quickly resume their VMs with all applications and processes intact, just as they were before.
Use cases
Specifically for GPU Virtual Machines, hibernation offers compelling use cases and is an effective cost management strategy particularly in two key scenarios:
- Optimizing GPU workstations: Pause GPU VMs during off-hours to conserve resources and resume seamlessly when needed, without the need to reopen applications.
- Efficient Workflows for long running VMs: For long-running GPU-intensive tasks, hibernating after prewarming tasks ensures quick start-up times and efficient use of GPU resources.
Getting started with Hibernation
Hibernation is generally available for general purpose Intel and AMD VM sizes, and on select sizes on GPU NVv4 and NVadsA10v5 series virtual machines. Linux and Windows Operating Systems are supported.
Hibernation is available in all public regions and customers using Azure Virtual Desktop, Citrix DaaS and Workspot on Azure can take advantage of hibernation today and further optimize costs. Get started with hibernation using Azure Portal, PowerShell, CLI, ARM, SDKs, and APIs. For more details on how to get started with hibernation, refer to the product documentation.
Microsoft