Analyze the provisioned IOPS and throughput for Hyperdisk volumes

You can view the disk description to see the provisioned input/outputoperations per second (IOPS) or the provisioned throughput forGoogle Cloud Hyperdisk volumes.

You can change the provisioned IOPS or throughput once every 6 hours forHyperdisk ML or once every 4 hours for all other Hyperdisk types.Each change of the IOPS or throughput level is logged. You can review thelog history and compare it with performance metrics to understand how theprovisioned IOPS and throughput levels relate to the performance levelobserved by your workload.

Before you begin

View the provisioned performance settings for Hyperdisk

To view the provisioned IOPS or throughput for your Hyperdiskvolumes, view the disk information.

Console

  1. In the Google Cloud console, go to theDisks page.

    Go to Disks

  2. Click the name of the disk to view the configuration details.

Screenshot of the configured properties for a Hyperdisk Balanced disk
Screenshot of the configured properties for a Hyperdisk

gcloud

  • Use thegcloud compute disks describe commandto view the disk details.

    gcloud compute disks describeDISK_NAME \  --zoneZONE_NAME \  --format="text(name, provisionedIops, provisionedThroughput, sizeGb)"

    Replace the following:

    • DISK_NAME: the name of theHyperdisk volume.
    • ZONE_NAME: the zone where theHyperdisk volume was created.

    The output shows the name of the disk, the current disk size and theprovisioned IOPS or throughput, for example:

    name: my-hyperdisk-bprovisionedIops: '8500'provisionedThroughput: '140'sizeGb: '150'

REST

Construct aGET request to thecompute.disks.get method.In the request body, specify the name of the Hyperdisk volume.

GET https://compute.googleapis.com/compute/v1/projects/PROJECT_ID/zones/ZONE/disks/DISK_NAME/get

Replace the following:

  • PROJECT_ID: your project ID.
  • ZONE: the zone where yourHyperdisk volume is located.
  • DISK_NAME: the name of theHyperdisk volume to view.

In the response body, you can view the current disk size, provisionedIOPS, and throughput, for example:

{  ...  "name": "my-hyperdisk-x",  "physicalBlockSizeBytes": "4096",  "provisionedIops": "100000",  ...  "sizeGb": "1000",  "status": "READY",  ...}

You can use a query filter to return only the information you want toview. To view only the fields shown in the preceding example output, appenda query parameter similar to the following to your request.

?fields=name,physicalBlockSizeBytes,provisionedIops,provisionedThroughput,sizeGb,status

View disk performance metrics

To view performance metrics for your VMs, use the Cloud Monitoringobservability metrics available in the Google Cloud console.

  1. In the Google Cloud console, go to theVM Instances page.

    Go to VM instances

  2. To view metrics for individual VMs:

    1. Click the name of the VM you want to view performance metrics for. The VMDetails page opens.

    2. Click theObservability tab to open the ObservabilityOverviewpage.

  3. Explore the VM's performance metrics. The following are key metricsrelated to disk performance for a VM:

    • On theOverview page:

      • CPU Utilization. The percent of CPU used by the VM.

      • Network Traffic. The average rate of bytes sent and receivedin one minute intervals.

      • Disk Throughput. The average rate of bytes written to and read fromdisks.

      • Disk IOPS. The average rate of I/O read and write operations todisks.

    • On theDisks Performance page, view the following charts:

      • Operations (IOPS). The average rate of I/O read and write operationsto the disk in one-minute time periods.

      • IOPS by Storage Type The average rate of I/O operations to the diskin one-minute time periods, grouped by the storage type and device type.

      • Throughput (MB/s) The average rate of bytes written to and read fromthe VM's disks in one-minute time periods.

      • Throughput by Storage Type The average rate of bytes written to andread from the VM's disks in one-minute time periods, grouped by thestorage type and device type.

      • I/O Size Avg. The average size of I/O read and write operations todisks. Small (4 to 16 KiB) random I/O operations are usuallylimited by IOPS and sequential or large (256 KiB-1 MiB)I/O operations are usually limited by throughput.

      • Queue Length Avg. The number of queued and running disk I/Ooperations, also calledqueue depth, for the top 5 devices. To reachthe performance limits of your Hyperdisk and Persistent Diskvolumes,use a high I/O queue depth.

      • I/O Latency Avg. The average latency of I/O read and write operationsaggregated across operations of all block storage devices attached to theVM, measured by the Ops Agent in the VM. This valueincludes operating system and file system processing time.

Analyze the IOPS needed for your workload

To determine the IOPS needed for your workload, make note of the peak andaverage IOPS and throughput rates during times of peak usage, and also during anormal workload cycle, to get an idea of your workload requirements.

Observe the IOPS requirements of your workload using any of the followingmethods:

  • Use theMonitoring tab on the disk details page in theGoogle Cloud console.
  • Use theObservability page for your VM, as described inView disk performance metrics.

Based on the observed metric values, determine if you should adjust theprovisioned IOPS for your VM. For example:

  • If the peak IOPS rate is close to the provisioned IOPS for theHyperdisk volume, then you can try increasing theprovisioned IOPS for the Hyperdisk volume to boost theperformance of your application.
  • If the peak IOPS rate is consistently lower than the provisioned IOPS, thenyou can lower the provisioned IOPS for the Hyperdisk volumeto reduce the cost of the disk.

Analyze the throughput needed for your workload

You can provision throughput separately from disk capacity for thefollowing Hyperdisk types:

  • Hyperdisk Balanced
  • Hyperdisk Balanced High Availability
  • Hyperdisk Throughput
  • Hyperdisk ML

You can specify the target throughput levelfor a given volume. Individual volumes have full performance isolation - eachvolume gets the performance provisioned to it. However, the throughput isultimately capped by per-VM limits for the VM to which yourvolumes are attached. To review these limits, seeHyperdisk performance limits.

Both read and write operations count against the throughput limit provisionedfor a Hyperdisk volume. The provisioned throughput and themaximum limits apply to the combined total of read and write throughput.

Observe the throughput requirements of your workload using any of the followingmethods:

  • Use theMonitoring tab on the disk details page in theGoogle Cloud console.
  • Use theObservability page for your VM, as described inView disk performance metrics.

If the total throughput provisioned for one or more Hyperdiskvolumes exceeds the total throughput available at the VM level,the performance is limited to the VM-level performance.

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.