Release Notes

This page contains the latest release notes for features and updates to the Compute Engine service. For older release notes, see the archive.

Latest API version: v1

To get the latest product updates delivered to you, add the URL of this page to your feed reader.

March 22, 2019

• Increased the per-instance persistent disk throughput performance to 240 Mb/s. See the Persistent Disk Performance for details.

March 11, 2019

• Added new europe-west6 region in Zürich, Switzerland. The europe-west6 region contains Skylake zones that are now available to all projects and users. See Regions and Zones for more information.

February 27, 2019

• Memory-optimized machine types with up to 160 vCPUs and 3.75 TB of memory are now available in the following zones:

  • asia-northeast1-a
  • asia-southeast1-b
  • northamerica-northeast1-b
  • northamerica-northeast1-c
  • southamerica-east1-b
  • southamerica-east1-c
  • us-central1-a

  See the pricing page to learn how these machine types are billed.

February 21, 2019

• You can now attach up to 128 independent persistent disks to custom machine types and most predefined machine types. For more details, see Persistent disk limits.

February 14, 2019

• User-created scheduled snapshots for zonal and regional persistent disks is now available in Beta. Scheduled snapshots allows you to create an automatic snapshot schedule and include a retention policy for maintaining those snapshots. For more information, see Creating scheduled snapshots for persistent disk.

February 11, 2019

• Sole tenant nodes are now available in Hong Kong (asia-east2).
• Application-based health checking and autohealing for managed instance groups is now Generally Available.

February 06, 2019

• Selecting 16 KB physical block size is now in Beta. You can create zonal and regional persistent disks with a 16 KB physical block. Some database applications might see greater performance by using 16 KB physical block. Read Adding Persistent Disks for feature details. Read Optimizing Persistent Disk and Local SSD Performance for performance comparisons. Read the Best practices for 16 KB persistent disk and MySQL tutorial to learn about performance improvements to a MySQL database.
• Users and applications can now write to instance-specific guest-writeable metadata values. You can use guest attributes to communicate status of applications and scripts within your instance. Read Setting and querying guest attributes to learn more.

February 6, 2019

• The Managed Instance Group Updater is now Generally Available. This feature enables proactive, flexible rolling updates with the option to canary a new version on a subset of managed instance group VMs.

January 29, 2019

• Granting access to specific Compute resources, such as VM instances, disks, and images, is now Generally Available. This feature gives you flexibility to apply the principle of least privilege, for example, to grant collaborators permissions only to the specific resources that they need to do their work.

January 24, 2019

• Detaching and reattaching boot disks from a stopped VM instance is now available in GA. Use the Google Cloud Platform Console, gcloud or API to detach a failed boot disk, fix it, then reattach it back to the original VM instance. You can also use this feature for rollbacks: detach a disk, create a new disk using an earlier snapshot, and attach that restored disk to your VM. See Detaching and Reattaching Boot Disks for details.

• Creating a VM Instance with a Custom Hostname is now Generally Available.

January 16, 2019

• NVIDIA® Tesla® T4 GPU and NVIDIA® Tesla® T4 Virtual Workstations are now available, as Beta, in eight regions. See the GPUs on Compute Engine page to learn more.

January 11, 2019

• Windows Server 2019 images are now Generally Available as a public image.
• Windows Server, version 1809 images are now Generally Available as a public image. This is a Server Core image that is part of the semi-annual release cycle for Windows Server.

December 14, 2018

• User-selected snapshot storage locations are now available in Beta. The storage location of a snapshot affects the availability of the snapshot and networking costs. For more information, see Selecting the storage location for a snapshot.
• Snapshot creation might now incur networking charges. These charges will begin after this feature reaches General Availability (GA), and will depend upon the storage location of a snapshot and the location of the source disk. For more information, see Network charges for snapshot creation and restoration.

November 15, 2018

• Windows Service Activation using Private Google Access is Generally Available. Read the Private Google Access page to learn more.

November 13, 2018

• Detaching and reattaching boot disks from a stopped VM instance is now available in Beta. Use the gcloud or API command to detach a failed boot disk, fix it, then reattach it to a different VM instance. See Detaching and Reattaching Boot Disks for details.

November 08, 2018

• Memory-optimized machine types with up to 160 vCPUs and 3.75 TB of memory are now available in the southamerica-east1-a zone. See the pricing page to learn how these machine types are billed.

November 07, 2018

• Creating a VM Instance with a Custom Hostname is available in Beta.

October 24, 2018

• Viewing VM serial port output in Stackdriver is available in Beta. To help with troubleshooting, these logs can be retained for up to 30 days after a VM instance is deleted.

October 22, 2018

• The asia-east2 Hong Kong region is now available to all projects and users. The zones in the asia-east2 region have the Skylake CPU platform. See Regions and Zones for more information.

October 19, 2018

• Memory-optimized machine types with up to 160 vCPUs and 3.75 TB of memory are now available in the asia-southeast1-c, australia-southeast1-c, europe-west3-a, europe-west3-b and us-west2-b zones. See the pricing page to learn how these machine types are billed.

October 02, 2018

• Compute Engine now uses a resource based pricing model that provides greater savings from sustained use discounts. All machine types except for shared-core machine types are now billed by their individual vCPU and memory usage rather than billing by each machine type. Sustained use discounts are calculated for vCPU and memory usage across an entire region rather than separately for each machine type in each zone. See the Compute Engine Pricing page for details.

• Protect data on Compute Engine with Cloud Key Management Service encryption keys. Customer-Managed Encryption Keys are now generally available.

September 26, 2018

• NVIDIA® Tesla® P4 GPUs are now Generally Available. For information about using GPUs on Compute Engine, see GPUs on Compute Engine.

• For graphics-intensive workloads, NVIDIA® GRID®-based virtual workstations using NVIDIA® Tesla® P4 and P100 GPUs are now Generally Available. For information on GPUs for graphics-intensive applications, see GPUs for graphics workloads.

• NVIDIA® Tesla® v100 GPUs are now available in the us-central1-b and europe-west4-b zones. See GPUs on Compute Engine for a complete list of zones where GPUs are available.

• NVIDIA® Tesla® P4 GPUs are now available in the australia-southeast1-a, and australia-southeast1-b zones. See GPUs on Compute Engine for a complete list of zones where GPUs are available.

September 25, 2018

• Windows Server instances no longer require a public IP for Windows Service Activation. If your Windows Server instance uses a subnetwork that is enabled for Private Google Access, that instance can complete Windows Service Activation over its internal VPC network. Windows Service Activation using Private Google Access is available in Beta. Read the Private Google Access page to learn more.

September 20, 2018

• Deploying Docker Containers on Compute Engine is now Generally Available.

September 12, 2018

• Now in Beta, you can grant access to specific Compute resources, such as VM instances, disks, and images. This feature gives you flexibility to apply the principle of least privilege, for example, to grant collaborators permissions only to the specific resources that they need to do their work.

September 06, 2018

• Zonal DNS names are now Generally Available. Projects that enable the Compute Engine API after today will use zonal DNS names by default. Projects and organizations that enabled the Compute Engine API before today will continue to use global DNS names by default. Migrating a project to an organization will not change the default DNS name for the project. Zonal DNS names are unique to each zone on your internal VPC networks. Zonal DNS names improve the fault tolerance of your applications when they reference instances on your internal VPC network. New and existing projects can still use global DNS names, but migration is encouraged. Read Internal DNS to learn more.

September 4, 2018

• The simulateMaintenanceEvent API method is Generally Available. Read testing your availability policies to learn how to simulate a Compute Engine maintenance event and observe how the instance availability settings affect the behavior of your instances.

August 27, 2018

• NVIDIA® Tesla® V100 GPUs are now Generally Available. For information about using V100 GPUs on Compute Engine, see GPUs on Compute Engine.

August 17, 2018

• Zonal SSD persistent disks and regional SSD persistent disks now have increased read throughput and read input/output operations per second (IOPS). See Optimizing Persistent Disk and Local SSD Performance for more information.

August 8, 2018

• Sole-tenant nodes are now Generally Available. Read Sole-tenant Nodes to learn how you can use these systems to isolate your workloads from each other and from other Compute Engine projects.

August 3, 2018

• If you have graphics-intensive workloads, such as 3D rendering or video editing, you can create virtual workstations that use NVIDIA® GRID® technology, using NVIDIA® Tesla® P4 and P100 GPUs. For information on GPUs for graphics-intensive applications, see GPUs for graphics workloads.
• You can use organization policies to restrict use of your disks, custom images, and snapshots so that they can be used only within your organization or specific projects. This prevents users from creating copies of disks, outside of your organization. Read restricting use of your shared images, disks, and snapshots to learn more.

July 25, 2018

• Shielded VM is now available in Beta. Shielded VM images offer security features like UEFI-compliant firmware, Secure Boot, and vTPM-protected Measured Boot.

July 23, 2018

• The n1-megamem-* family of memory-optimized machine types with 1.4 TB of memory are now Generally Available to all projects. See the pricing page to learn how these machine types are billed.

July 19, 2018

• Selecting specific zones for regional managed instance groups is now Generally Available. See Distributing Instances using Regional Managed Instance Groups.

• Windows Server for Containers is now Generally Available as a public image.

July 18, 2018

• Memory-optimized machine types with up to 160 vCPUs and 3.75 TB of memory are now Generally Available. See the pricing page to learn how these machine types are billed.

July 17, 2018

• If you need high availability and redundancy for your Microsoft SQL Server on Compute Engine, you can configure an Always On Failover Cluster Instance (FCI) with Storage Spaces Direct (S2D).

July 11, 2018

• Virtual machine sizing recommendations are now Generally Available.

July 10, 2018

• Added new us-west2 Los Angeles region. us-west2 contains Skylake zones that are now available to all projects and users. See Regions and Zones for more information.

• If you try to create a new instance in this zone for a project that existed before this release date, you may receive the following error message: "No default subnetwork was found in the region of the instance." This issue will be resolved within three days of this launch. In the meantime, the workaround is to manually create a subnet and use it for a new VM.

• Creating an instance template from a VM instance is now Generally Available. This feature can be used for instance duplication, backup, and recreation.

July 2, 2018

• You can now add 2 and 4 NVIDIA® Tesla® V100 GPUs to your Compute Engine instances. For information about using GPUs on Compute Engine, see GPUs on Compute Engine.

June 22, 2018

• You can purchase committed use discounts for memory-optimized machine types. Memory-optimized machine types use their own commitment type with different pricing. You must request access to use committed use discounts for memory-optimized machine types. See the pricing page for pricing details about committed use discounts.

June 19, 2018

• The Compute Engine Trusted Images policy is now Generally Available. Use this policy to control which boot disk images your project members can access. Read Restricting Access to Images for more information.

June 11, 2018

• Added new europe-north1 region. europe-north1 contains Skylake zones that are now available to all projects and users. See Regions and Zones for more information.

• If you try to create a new instance in this zone for a project that existed before this release date, you may receive the following error message: "No default subnetwork was found in the region of the instance." This issue will be resolved within three days of this launch. In the meanwhile, the workaround is to manually create a subnet and use it for a new VM.

• The prices for GPUs on preemptible VM instances have reduced. For the updated prices, see the Compute Engine pricing page.

• GPUs on preemptible VM instances are now Generally Available.

  For information about using GPUs on Compute Engine, see GPUs on Compute Engine.

June 07, 2018

• You can create sole-tenant nodes, which are physical Compute Engine servers that are dedicated to hosting only VM instances from your specific project. Read Sole-tenant Nodes to learn how you can use these systems to isolate your workloads from each other and from other Compute Engine projects.

June 05, 2018

• Protect data on Compute Engine with Cloud Key Management Service encryption keys. Customer-Managed Encryption Keys are now available in beta.

May 28, 2018

NVIDIA® Tesla® V100 GPUs are now available in the following regions:

• us-west1-a
• us-central1-a
• asia-east1-c

For information about using V100 GPUs on Compute Engine, see GPUs on Compute Engine.

May 23, 2018

• Added support for regional persistent disks into Beta. This feature provides synchronous block-level replication across two zones in a region for each regional persistent disk.

May 16, 2018

• Instances with NVIDIA® Tesla® V100 GPUs can now use Local SSD devices. For information on using V100 GPUs on Compute Engine, see GPUs on Compute Engine.

May 08, 2018

• Memory-optimized machine types with up to 160 vCPUs and 3.75 TB of memory are now available in Beta. See the pricing page to learn how these machine types are billed.

May 07, 2018

• Added local SSD support to the europe-west4-a zone. See Regions and Zones for more information.

May 01, 2018

• Added a third zone to the asia-southeast1 region. This zone supports machine types with up to 96 vCPUs when using the Skylake platform. See Regions and Zones for more information.

April 30, 2018

• NVIDIA® Tesla® V100 GPUs are now available in Beta. For information on the zones where V100 GPUs are available, see GPUs on Compute Engine.

April 20, 2018

• NVIDIA® Tesla® P100 GPUs are now Generally Available. To learn about the zones where P100 GPUs are available, read GPUs on Compute Engine.

• Creating multiple secondary non-boot disks when you create your VM instance is now Generally Available.

April 19, 2018

• The price of SLES for SAP images is reduced. Read SUSE premium image pricing for details.

April 17, 2018

NVIDIA® Tesla® P100 GPUs are now available in the following zones:

• asia-east1-c
• europe-west4-a

To learn more about using GPUs on Compute Engine, see GPUs on Compute Engine.

April 11, 2018

• You can now run simulated maintenance events to test the effects of live migration, termination, and preemption on your instances. Simulated maintenance events are available in Beta. Read testing your availability policies to learn how to test maintenance events on your instances.

• Nested Virtualization is now Generally Available.

March 28, 2018

• You can import existing virtual disks (VMDKs, VHDs, etc) into Compute Engine as custom images. Importing virtual disks is now Generally Available.

March 23, 2018

• You can import a Windows Server 2003 server to Compute Engine using the CloudEndure migration service. For steps, see Migrating VMs to Compute Engine.

March 14, 2018

• Creating an image from a snapshot is now Generally Available. See the creating custom images page.

March 13, 2018

• Added a third zone to the europe-west4 region. This zone supports 96 vCPU high memory/high CPU machines. There is no support for local SSDs at this time. See Regions and Zones for more information. For regional managed instance group users, the addition of a third zone in europe-west4 offers you more flexibilty to specify zones in this region. Note that if you do not specify the zones for your instances, Compute Engine, by default, selects all three zones. See Distributing Instances using Regional Managed Instance Groups to learn more.

March 08, 2018

• Added support for PTR records on VM instances to General Availability. See Creating a PTR record for VM instances.

March 07, 2018

• You can import existing virtual disks (VMDKs, VHDs, etc) into Compute Engine as functioning images. Importing virtual disks is available in Beta.

March 02, 2018

• Mega-memory machine types with 1.4 TB of memory are now available in Beta. See the pricing page to learn how these machine types are billed.

February 22, 2018

• Creating multiple secondary non-boot disks when you create your VM instance is now available in Beta.

February 9, 2018

• 96 vCPU machine types are Generally Available. These machine types are now available in several additional zones only on the Skylake platform. Read the list of available regions and zones to see where 96 vCPU machine types are available.

February 1, 2018

• Added support for creating an instance template from a VM instance into Beta. This feature can be used for instance duplication, backup, and recreation.

January 31, 2018

• Compute Engine supports IAM Custom Roles. Read the IAM Custom Roles documentation to learn more. You can identify which permissions are required to run Compute Engine methods in both the V1 Compute Engine API reference and on the Compute Engine IAM Permissions page.

• You can configure autoscaling for groups of instances based on a wider range of metrics, which allow you to scale managed instance groups on per-group metrics rather than average resource utilization across all of the instances in a group. Autoscaling using per-group metrics is available in Beta.

• You can filter Stackdriver monitoring metrics when you use them to configure autoscaling for your managed instance groups. Filtering per-instance metrics is available in Beta.

January 24, 2018

• The OS Login API and key management feature is now Generally Available. You can associate your public SSH keys with your Google account or with individual member accounts in a G Suite organization or Cloud Identity organization. Read Managing Instance Access for more information.

January 22, 2018

• For NVIDIA® Tesla® K80 GPUs in the asia-east1-a and us-east1-d zones, you can create GPU instances with up to 416 GB of memory.

To learn more about the zones where GPUs are available, read GPUs on Compute Engine.

January 11, 2018

• VM Deletion Protection is now Generally Available.

January 10, 2018

• Added new europe-west4 region. europe-west4 contains Skylake zones that are now available to all projects and users. See Regions and Zones for more information.

• Added new northamerica-northeast1 Montréal region. northamerica-northeast1 contains Skylake zones that are now available to all projects and users. See Regions and Zones for more information.

January 8, 2018

• Zonal DNS names are available in Beta. Zonal DNS names are unique to each zone on your internal VPC networks. Global DNS names continue to resolve, but you can enable zonal DNS names to improve the fault tolerance of your applications when they reference instances on your internal VPC network. Read Zonal DNS to learn more.