100% Free Forever
AI-Powered Learning
Industry Expert Content
Certificates & Badges
Learn At Your Own Pace
Azure

Azure Disks Explained

How Azure Managed Disks work, including performance tiers, OS/data/temporary disk roles, caching, and snapshotting.

StorageIntermediate10 min readJul 10, 2026
Analogies

What Are Azure Managed Disks?

Azure Managed Disks are block-level storage volumes attached to virtual machines, presented to the VM as if they were physical hard drives. Azure handles the underlying storage accounts automatically, so you no longer need to manage storage account capacity limits or manually place VHDs to avoid IOPS throttling, which was a common pain point with the older unmanaged disk model. Each managed disk is a standalone Azure resource with its own resource ID, enabling features like fine-grained access control, snapshots, and disk encryption.

🏏

Cricket analogy: Managed disks are like a franchise using a professional groundstaff contracted by the board to prep every pitch to spec, instead of the team captain personally worrying about soil composition and drainage before each match.

Disk Types: HDD, SSD, Premium SSD, Ultra Disk

Azure offers four managed disk performance tiers. Standard HDD is the cheapest, suited to backup or infrequently accessed data with tolerant latency requirements. Standard SSD offers better latency and reliability than HDD at moderate cost, good for web servers with light-to-moderate traffic. Premium SSD delivers low-latency, high-IOPS performance for production databases and I/O-intensive workloads, and requires a VM size that supports premium storage. Ultra Disk offers the highest performance tier, with configurable IOPS and throughput independent of disk size, ideal for the most demanding workloads like SAP HANA.

🏏

Cricket analogy: Standard HDD is like a club-level pitch adequate for a friendly match; Standard SSD is a decent domestic ground; Premium SSD is an international stadium wicket; Ultra Disk is a purpose-built World Cup final pitch tuned to exact specifications on demand.

bash
# Create a Premium SSD managed data disk and attach it to a VM
az disk create \
  --resource-group rg-disks-demo \
  --name db-data-disk01 \
  --size-gb 512 \
  --sku Premium_LRS \
  --location eastus

az vm disk attach \
  --resource-group rg-disks-demo \
  --vm-name db-server-01 \
  --name db-data-disk01

# Create an Ultra Disk with configurable IOPS and throughput
az disk create \
  --resource-group rg-disks-demo \
  --name sap-hana-disk \
  --size-gb 1024 \
  --sku UltraSSD_LRS \
  --disk-iops-read-write 20000 \
  --disk-mbps-read-write 400 \
  --location eastus \
  --zone 1

OS Disks vs Data Disks vs Temporary Disks

Every VM has an OS disk containing the boot volume and operating system, one or more optional data disks for application data and databases, and a temporary disk (labeled D: on Windows or /dev/sdb on Linux) that is local, ephemeral, non-persisted storage provided free with the VM size for page files, swap space, or scratch data — data on the temporary disk is lost on deallocation or migration. You can also configure disk caching (None, ReadOnly, or ReadWrite) per attached disk to improve performance for specific I/O patterns.

🏏

Cricket analogy: The OS disk is like the team's permanent training ground; data disks are like additional practice nets rented for a specific tour; the temporary disk is like a hotel room used only during the tour that's cleared out once the team departs.

For read-heavy database workloads on data disks, enabling ReadOnly caching can significantly reduce latency by serving repeated reads from the host cache instead of round-tripping to the underlying storage. Avoid ReadWrite caching on disks used for write-heavy or crash-sensitive workloads, since cached writes may not be as durable.

Disk Snapshots and Resizing

A disk snapshot is a full, point-in-time, read-only copy of a managed disk that you can use to create a new disk, restore a VM, or migrate a workload to another region. Incremental snapshots only store changed blocks since the last snapshot, reducing storage costs for frequent backups. Resizing a managed disk to a larger size is straightforward for data disks that can be resized while detached, but resizing an OS disk or a disk still attached to a running VM typically requires deallocating the VM first, which causes downtime.

🏏

Cricket analogy: A snapshot is like freezing the exact scorecard and field placements at the fall of a wicket so the situation can be replayed later; resizing the ground mid-match, however, requires pausing play entirely.

Resizing an OS disk or an attached data disk on a running production VM usually forces a deallocation, meaning the VM stops and loses its dynamic (ephemeral) public IP unless a static IP is configured. Always plan disk resize operations during a maintenance window.

  • Managed disks are block-level storage attached to VMs, with Azure automatically handling the underlying storage infrastructure.
  • Four performance tiers exist: Standard HDD, Standard SSD, Premium SSD, and Ultra Disk, ranging from cheapest to highest-performance.
  • Ultra Disk allows independently configurable IOPS and throughput, ideal for the most demanding workloads like SAP HANA.
  • Every VM has an OS disk, optional data disks, and a local, ephemeral temporary disk that is wiped on deallocation.
  • Disk caching (None/ReadOnly/ReadWrite) can be tuned per disk to optimize latency for specific I/O patterns.
  • Snapshots create point-in-time, read-only copies of disks; incremental snapshots reduce cost by storing only changed blocks.
  • Resizing an OS disk or an attached disk typically requires VM deallocation, so plan for a maintenance window.

Practice what you learned

Was this page helpful?

Topics covered

#Azure#AzureFundamentalsStudyNotes#CloudComputing#AzureDisksExplained#Disks#Explained#Managed#Disk#StudyNotes#SkillVeris