About Hyperdisk Extreme

This document describes the features and performance limits of Hyperdisk Extremevolumes. Hyperdisk Extreme is designed for applications that require more than160,000 IOPS or 2,400 MiB/s of bandwidth from a single volume, such ashigh-performance databases.

You can specify up to 350,000IOPS for a single Hyperdisk Extreme volume. You can't provision a throughput level forHyperdisk Extreme volumes. Each volume gets 250 MiB/s of throughputwith every 1,000 IOPS, up to 5,000 MiB/s.

Hyperdisk Extreme volumes are designed for sub-millisecond latencies.

Use cases

Hyperdisk Extreme is a good fit for the following use cases:

  • SAP HANA
  • High-end SQL Server, Oracle
  • In-memory RDBMS that require more than 160,000 IOPS

Machine series support

You can use Hyperdisk Extreme with the following machine series:

Restrictions for machine series support

  • C3 machine type require at least 88 vCPUs.
  • C3D machine types require at least 60 vCPUs.
  • C4 and G4 machine types require at least 96 vCPUs.
  • M1 machine types require at least 80 vCPUs.
  • C4A, C4D, M3, and M4 machine types require at least 64 vCPUs.
  • N2 machine types require at least 80 vCPUs; Custom N2 machine types aren't supported

About provisioned performance

You don't have to provision performance when you createHyperdisk volumes. If you don't provision performance, Compute Enginecreates the volume with default values that you can modify later.For details about default values, seeDefault IOPS andthroughput values.

You can specify an IOPS limits for a Hyperdisk Extreme volume when you create the volume,and you can modify the provisioned IOPS after you create the volume.You can't specify an IOPS level if you don't specify a size.

Important: Hyperdisk volumescan't reach the provisioned performance unless the compute instance supports thatlevel of performance. For detailed performance limits for all supported instancesby machine type, seePerformance limits when attached toan instance.

For details about performance for Hyperdisk volumes, seeAbout Hyperdisk performance.

Size and performance limits

The following limits apply to the size, throughput, and IOPS values you canspecify for a Hyperdisk Extreme volume.

  • Size: between 64 GiB and 64 TiB. The defaultsize is 1 TiB.
  • IOPS: between 128 and 350,000 IOPS.However, for volumes less than 1 TiB, the range of IOPS values you canspecify varies by size. For examples,seeLimits for provisioned IOPS.
  • Throughput: you can't specify a throughput limit for Hyperdisk Extreme volumes.Instead, the throughput depends on the provisioned IOPS. Each Hyperdisk Extreme volumeis provisioned with 250 MiB/s of throughput for every1,000 IOPS, up to a maximum of 5,000 MiB/s.

Limits for provisioned IOPS

The IOPS level you can specify for a Hyperdisk Extreme volume depends on the disk's size,according to the following formulas, wherex is the volume's size in GiB.

  • Minimum IOPS:2x
  • Maximum IOPS:
    • 64 GiB to 291 GiB:1,200x
    • 292 GiB to 64 TiB: 350,000 IOPS

This table provides the range of IOPS values that you can provision for commonvolume sizes. If a size value isn't listed, use the preceding formula tocalculate the maximum allowable IOPS.

Volume size (GiB)Minimum IOPSMaximum IOPS
6412876,800
150300180,000
200400240,000
290580348,000
300600350,000
1,0002,000350,000

Default size, IOPS, and throughput values

If you don't specify a size or IOPS value when you create a Hyperdisk Extremevolume, Compute Engine provisions the volume with default values.

The default size is 1 TiB.

The default IOPS and throughput are based on the following formulas.

  • Default IOPS: 100,000 IOPS if you don't specify asize. Otherwise,100x, wherex is thevolume's size in GiB.
  • Default throughput: 5,000 MiB/s if you don't specify anIOPS value. Otherwise, 250 MiB/s per 1,000 IOPS, up to5,000 MiB/s.

Change the provisioned IOPS or size

You can change the provisioned performance and size of a Hyperdisk Extreme volume at mostonce every 4 hours. For instructions on modifying sizeor performance, seeModify a Hyperdisk volume.

Performance limits when attached to an instance

This section lists the specific performance limits for each machine type thatsupports Hyperdisk Extreme. You can specify up to 350,000IOPS for a single Hyperdisk Extreme volume. You can't provision a throughput level forHyperdisk Extreme volumes. Each volume gets 250 MiB/s of throughputwith every 1,000 IOPS, up to 5,000 MiB/s.

This section lists the maximum performance that Hyperdisk Extreme volumescan achieve for each supported instance. A Hyperdisk Extreme volume's performancewhen it's attached to an instance can't exceed the limits for the instance'smachine type.The performance limits are also shared across all Hyperdisk Extreme volumesattached to the same instance, regardless of each volume's provisioned performance.

Achieve higher performance with multiple Hyperdisk Extreme volumes

Certain instances can exceed the maximum performance for a single Hyperdisk Extreme volume—350,000 IOPS or 5,000  MiB/s. An instance can only achieve thesehigher limits if you attach multiple Hyperdisk Extreme volumes to the instance.

If an instance's performance limit in the following table exceeds 350,000 IOPS or5,000  MiB/s, then the instance can only achieve that limitif you attach multiple Hyperdisk Extreme volumes.For example, the performance limit for ac3-*-176 instance using Hyperdisk Extremeis 500,000 IOPS and 10,000 MiB/s.To achieve this maximum performance you must attach at least two Hyperdisk Extreme volumes to the instance.

Important: Hyperdisk performance is half-duplex,so the IOPS and throughput limits are shared between read and write operations.This means that if a machine type has a 50,000 IOPS limit, then the sum of the readsand writes every second can't exceed 50,000. To reach the IOPS limits, you must use an I/O size of4 KB. Maximum throughput limits require an I/O size of at least256 KB.

Instance machine typeMaximum IOPSMaximum throughput (MiB/s)
A3 (A3+H100)
a3-highgpu-4g350,0005,000
a3-megagpu-8g400,0008,000
a3-highgpu-8g400,0008,000
a3-edgegpu-8g400,0008,000
A3 (A3+H200)
a3-ultragpu-8g2500,00010,000
A42
a4-highgpu-8g2500,00010,000
A4X2
a4x-highgpu-4g2500,00010,000
C31
c3-*-881350,0005,000
c3-*-1761500,00010,000
c3-*-1921,3500,00010,000
C3D
c3d-*-601350,0005,000
c3d-*-901350,0005,000
c3d-*-1801350,0005,000
c3d-*-3601350,0005,000
C42
c4-*-962350,0005,000
c4-*-1442350,0005,000
c4-*-1922500,00010,000
c4-*-2882,3500,00010,000
C4A
c4a-*-64350,0005,000
c4a-*-72350,0005,000
C4D2
c4d-*-642350,0005,000
c4d-*-962350,0005,000
c4d-*-1922350,0005,000
c4d-*-3842500,00010,000
G4
g4-standard-96350,0005,000
g4-standard-192350,0005,000
g4-standard-384500,00010,000
M2
m2-*-208100,0002,400
m2-*-416100,0002,400
M31
m3-*-641350,0005,000
m3-*-1281450,0007,200
M42
m4-*-642200,0003,000
m4-*-1122350,0005,000
m4-*-2242500,00010,000
N2
n2-*-801160,0005,000
n2-*-96160,0005,000
n2-*-128160,0005,000
X4
x4-480-*-metal1400,00010,000
x4-960-*-metal1400,00010,000
x4-1440-24t-metal1400,00010,000
x4-1920-32t-metal1400,00010,000
Z31
z3-*-881350,0005,000
z3-*-1761350,0005,000
z3-*-1921,3500,00010,000

1If using Hyperdisk Extreme with an instance that uses Microsoft Windows, refer to theknown issues for Windows VM instances.
2Also offers steady state performance. For more information, seePerformance limits for machine types with steady state performance.
3Includes bare metal instances.

Share a Hyperdisk Extreme volume between VMs

You can use multi-writer mode to simultaneously attach a Hyperdisk Extreme volume tomultiple instances, while each instance maintains write access. This featureenables high availability for workloads like Microsoft SQL Server Failover ClusterInfrastructure.

To learn more, seeShare a disk between VMs.

Disaster protection for Hyperdisk Extreme volumes

You can back up a Hyperdisk Extreme volume withinstant snapshotsandstandard snapshots.Snapshots back up the data on a Hyperdisk Extreme volume at a specific point in time.

You can protect your data in the unlikely event of a regional outage by enabling Asynchronous Replication. Asynchronous Replication maintains a copy ofthe data on your volume in another region. For example, to protect aHyperdisk Extreme volume inus-west1,you can use Asynchronous Replication to replicate the volume to a secondary volume intheus-east4 region. If the volume inus-west1became unavailable, then you could use the secondary volume inus-east4.

Cross-zonal replication

You can't replicate Hyperdisk Extremevolumes to another zone. To replicate data to another zone within the same region,you must useHyperdisk Balanced High Availability volumes.

Regional availability for Hyperdisk Extreme

Hyperdisk Extreme is available in all zones and regions.

Limitations

Hyperdisk Extreme volumes have the following limitations:

  • Hyperdisk Extreme volumes are zonal and can be accessed from only thezone where they are located.
  • Hyperdisk Extreme volumes can only be attached to machines with vCPUsof 60 or higher.
  • You can't create an image from a Hyperdisk Extreme volume.
  • You can't create a machine image from a Hyperdisk Extreme volume.
  • You can't use a Hyperdisk Extreme volume as a boot disk.
  • You can't attach Hyperdisk Extreme volumes to more than one VM.
  • You can change a Hyperdisk Extreme volume's size and performance every 4hours.

  • Windows instances running onn2-standard-80 machine types can reach at most80,000 IOPS across all Hyperdisk Extreme volumes that are attached to the instance.To reach up to 160,000 IOPS with N2 instances running Windows, choose oneof the following machine types:

    • n2-highmem-80
    • n2-highcpu-80
    • n2-standard-96
    • n2-highmem-96
    • n2-highcpu-96
    • n2-highmem-128
    • n2-standard-128

What's next

Except as otherwise noted, the content of this page is licensed under theCreative Commons Attribution 4.0 License, and code samples are licensed under theApache 2.0 License. For details, see theGoogle Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.

Last updated 2025-12-15 UTC.