DEV Community

Latchu@DevOps
Latchu@DevOps

Posted on

Part-27: 🚀 Google Cloud Hyperdisk and Storage Pools – A Hands-On Guide

#googlecloud #gcp #cloud #tutorial

Google Cloud keeps innovating in the storage space for Compute Engine. While Persistent Disks (PDs) have been the default choice for years, Google recently introduced Hyperdisk – a next-gen block storage solution.

In this post, we’ll break down:

🔹 What Hyperdisk is and how it differs from Persistent Disks

🔹 Types of Hyperdisks and their use cases

🔹 Key limitations to watch out for

🔹 What Storage Pools are and why they matter

🔹 Step-by-step demo: Create a Hyperdisk & explore Storage Pool options

🔹 Hyperdisk – Next-Gen Block Storage

Hyperdisk is the newest generation of network block storage in Google Compute Engine.

✅ Key Difference: Persistent Disk vs Hyperdisk

  • Persistent Disk (PD): Performance scales automatically with disk size + VM CPU.
  • Hyperdisk: You can directly provision performance (IOPS & throughput) independent of size.

This makes Hyperdisk more flexible for performance-sensitive workloads.

👉 Hyperdisk also offers:

  • Dedicated IOPS & throughput guarantees
  • NVMe or SCSI interface options
  • Same lifecycle behavior as PD (attach/detach, persistence across reboots/deletes)

🔹 Types of Hyperdisks

Google currently offers three Hyperdisk types, each tuned for different workloads:

Hyperdisk Balanced

  • General-purpose storage
  • Use cases: Web applications, medium-tier databases

Hyperdisk Extreme

  • For performance-critical workloads where even Extreme PD isn’t enough
  • Use cases: High-performance databases

Hyperdisk Throughput

  • Designed for workloads needing high throughput at scale
  • Use cases: Hadoop, Kafka, scale-out analytics, cost-sensitive data processing

🔹 Hyperdisk Limitations

Before diving in, be aware:

❌ No images or machine images can be created from Hyperdisk Extreme / Throughput volumes

❌ You cannot clone a Hyperdisk volume

❌ Hyperdisk is zonal only (no regional Hyperdisks)

❌ No multi-attach, no read-only sharing across multiple VMs

❌ Extreme / Throughput cannot be used as boot disks

🔹 Hands-On: Create a Hyperdisk

Step 01: Introduction

We’ll create a Hyperdisk Balanced volume and review its properties.

Step 02: Create Hyperdisk (Console)

Go to Compute Engine → Storage → Disks → Create Disk
Fill in details:

  • Name: hyperdisk1
  • Type: Hyperdisk Balanced
  • Size: 100 GB
  • Provisioned IOPS: 3600 (default)
  • Provisioned Throughput: 290 (default)
  • Zone: us-central1-a

Click Create

Step 02 (CLI):

# Create Hyperdisk Balanced
gcloud compute disks create hyperdisk2 \
    --project=gcplearn9 \
    --type=hyperdisk-balanced \
    --description=hyperdisk1 \
    --size=100GB \
    --zone=us-central1-a \
    --provisioned-iops=3600 \
    --provisioned-throughput=290
Enter fullscreen mode Exit fullscreen mode

Step 03: Review Disk

Go to Compute Engine → Storage → Disks → hyperdisk1 to confirm properties.

Step 04: Cleanup

gcloud compute disks delete hyperdisk1 --zone us-central1-a
Enter fullscreen mode Exit fullscreen mode

🔹 Storage Pools – Simplifying Large-Scale Storage

When managing hundreds or thousands of disks, it becomes painful to manually size capacity, IOPS, and throughput.

That’s where Storage Pools come in:

  • Pre-purchased capacity + IOPS + throughput pool
  • Allocate resources to VMs dynamically
  • Designed for large-scale, SAN-like workloads

✅ Key Features of Storage Pools

  • Capacity Thin Provisioning → allocate blocks as needed, not upfront
  • Data Reduction → compression increases storage efficiency
  • Auto-grow → when pool reaches 80% usage, GCP auto-expands it

🔹 When to Use Storage Pools?

  • Migrating workloads from on-prem SAN → no need to overprovision upfront
  • Underutilization of block storage resources → pools auto-manage growth
  • Complex environments with hundreds of disks → pools reduce management effort

🔹 Storage Pool Limitations

  • Zonal only (no regional pools)
  • Only works with Compute Engine, not Cloud SQL
  • Max: 1 PiB capacity & 1000 disks per pool
  • No instant snapshots or cloning
  • Only supports new disks in same project & zone

🔹 Hands-On: Create Storage Pool

⚠️ Note: Creating storage pools incurs significant cost, so here we’ll just review options.

Step 01: Create Storage Pool (Console)

  • Go to Compute Engine → Storage → Storage Pools
  • Name: storage-pool-1
  • Type: Hyperdisk Balanced
  • Capacity: 10 GB
  • IOPS: 10,000
  • Throughput: 1024
  • Submit

Step 01 (CLI):

# Create Storage Pool
gcloud compute storage-pools create storage-pool-1 \
  --project=gcplearn9 \
  --provisioned-capacity=10240 \
  --storage-pool-type=hyperdisk-balanced \
  --zone=projects/gcplearn9/zones/us-central1-a \
  --provisioned-iops=10000 \
  --provisioned-throughput=1024
Enter fullscreen mode Exit fullscreen mode

🔹 Final Thoughts

  • Hyperdisk brings flexible, provisioned performance that PDs can’t match.
  • Storage Pools simplify resource planning and management for large-scale deployments.
  • Use Balanced Hyperdisk for general workloads, Extreme for high-performance databases, and Throughput for analytics pipelines.
  • Always review limitations (zonal only, no cloning, limited boot disk support) before adopting.

With this, you now have a solid understanding of Hyperdisk & Storage Pools in GCP 🚀

Top comments (0)