This page provides an overview of bringing your own license (BYOL) to Compute Engine using sole-tenant nodes and the in-place restart feature. Before getting started, review your licensing terms and conditions to confirm that you meet the licensing requirements to bring your own license to Compute Engine. For step-by-step instructions, read the how-to guide to bring your own license.
If you have existing operating system licenses, you can bring them to Compute Engine using sole-tenant nodes while minimizing physical core usage with the in-place restart feature. Sole-tenant nodes are physical Compute Engine servers that are dedicated to hosting VM instances only for your specific project. When you enable in-place restart on sole-tenant nodes, Compute Engine minimizes the number of physical servers your VM jobs by restarting the VM on the same server whenever possible. If restarting VMs on the same physical server is not possible (for example, if the physical server experiences a critical hardware failure), your VMs are migrated to another server. Compute Engine assigns and reports a new physical server ID, and the old server ID is permanently retired.
This is especially useful during host maintenance events. Instead of live migrating to a new physical server, Compute Engine terminates and restarts the VM on the same server. VMs are taken offline and are unavailable while maintenance is applied.
Hosting instances on sole-tenant nodes provides the following benefits:
- Your VMs run on hardware that is exclusively dedicated to your project.
- The number of physical servers your VM instances will touch during maintenance events is minimized.
- You can track per-socket and per-core usage, information that might be required for license reporting purposes.
- You control node placement, so that your VMs can be scheduled on a specific sole tenant node or scheduled across a set of matching nodes.
- Compute Engine reports when instances are created or deleted on a specific server using a server ID, which supports server usage reporting. You can combine this information with the server's physical characteristics to determine information for license usage.
- You can bring your own licenses for Windows Server and Windows client.
For Windows apps such as SharePoint Server, Microsoft Exchange Server, and so on, existing app licenses can be deployed on Google Cloud using license mobility instead. See the frequently asked questions for details.
Compatible OS versions
The current Compute Engine image import workflow has been tested for compatibility with the following OS versions:
- Windows Server (64-bit only): 2012, 2012 R2, 2016, 2019
- Windows Client (32-bit and 64-bit): 7 SP1, 8.1, 10 (versions 1709, 1803, 1903, 1909)
If you would like to import Windows 7 SP1 or Windows 2008 R2 SP1, the image must support SHA2 signing since Compute Engine drivers are signed with SHA2. Also note that certain agents from 3rd party cloud providers might interfere with the Compute Engine driver installation workflow and should be removed prior to import.
Pricing and availability
This feature is available in all regions that support sole tenant nodes. To view the complete list, see Regions and zones.
You aren't charged extra for bringing your own licenses to Compute Engine.
Licensing and activation
Before getting started with BYOL with sole tenant nodes, make sure that you understand your product licensing terms.
You must also prepare your guest image for BYOL according to your agreements. If you use Compute Engine public images, Google Cloud pay-as-you-go licenses are attached to VM instances that use these images and you can't update the VM to use your own licenses later.
Unlike premium images where Compute Engine takes care of license activation, imported images require that you own the license activation process. You can't activate licenses against the Compute Engine license server. You must prepare infrastructure for activation, such as a KMS server and network connection, and activating your licenses.
Before attempting to bring your own license, we strongly recommend that you review the following information to help you determine the best workload for this feature.
You can use this feature to enable in-place restarts for your VM instances running on sole tenant nodes. This means that, during maintenance events, instead of live migrating a running instance to another machine and exposing your license to another physical server, the instance is terminated and restarted on the same physical server whenever possible. If that server is no longer available, Google restarts your VMs on a new dedicated server and reports the server ID.
Google performs an average of one maintenance event a month, which means the VMs on your server will experience an outage while maintenance is applied. Each event lasts about 60 minutes each, though this could vary depending on the situation, such as type of maintenance, whether it is a patching event, and so on.
Because of this expected downtime, we recommend using workloads that are tolerant of the downtime associated with these maintenance events.
Physical server migration
On rare occasions, Compute Engine might need to retire a physical
server and move your VMs to a new underlying server due to an event such as a
critical hardware failure. During these situations, Compute Engine
restarts your VMs on a new physical server and assigns a new sole tenant
physical server ID. If you set the
--restart-on-failure VM property, the
VM is restarted on the newly provisioned physical server. After this happens,
the previous sole tenant physical server ID isn't reused.