Azure Premium Storage

In December 2014, Microsoft announced the Public Preview Availability release of Azure Premium Storage. Here are the features of Azure Premium Storage (as of January 2015):

Limited Public Preview:

This means you have to sign up for it, and wait to get approved before you can use it. AzurePS01

We also must use the new Azure Portal to manage it. It’s available in West US, East US 2, and West Europe only.

Premium Storage account

This is required for Azure Premium Storage. It has a 32 TB capacity limit, 50K IOPS limit, and can contain Page blobs only.


On the other hand a Standard Storage account has a 500 TB capacity limit, 20k IOPS limit, and can contain both page blobs and block blobs.


This means it’s available to the Azure VM as it live-migrates from one Hyper-V host to another.

This is unlike the A-Series Azure VM drive ‘d’ which is SAS local temporary storage on the Hyper-V host. It provides a ‘scratch volume’ where we can take advantage of IO operations but data can be lost as the VM is live-migrated from one host to another during planned Azure maintenance or unplanned host failure. The D-Series and DS-Series Azure VMs have SSD local temporary storage drive ‘d’.


Solid State Disk based, as opposed to Azure Standard Storage which is SAS spinning hard disk based.

Available to DS series VMs only:

So, we can provision Premium Storage as Page blobs (not block blobs), with the current limitations (maximum size per file is 1,023 GB, must be VHD format not VHDX). Each page blob is a VHD disk file that can be attached to an Azure VM.

Maximum 32 disks per VM:

We can attach up to 32 of these data disks to a single DS-Series VM. For example, using the following Powershell cmdlet:

Get-AzureRoleSize | where { $_.MaxDataDiskCount -gt 15 } |
    Select RoleSizeLabel,Cores,MemoryInMB,MaxDataDiskCount | Out-GridView



Locally Redundant. 3 copies of each block are maintained synchronously in the same Azure facility.

Size, Performance:


  • IO/s are calculated based on 256KB block size.
  • Throughput limits include both read and write operations. So, for a P10 it’s 100MB/s for both read and write operations combined, not each of read and write operations
  • Exceeding disk type maximum IO/s or Throughput will result in throttling, where the VM will appear to freeze.
  • Less than 1 ms latency for read operations
  • Read-caching is set by default on each disk but can be changed to None or Read-Write using Powershell.


The only VM size that allows 32 TB of data disk space is the Standard_D14, which currently costs $,1765/month


Each 1TB Premium Storage disk costs $74/month


This adds up to almost $50k/year for a single VM


That would be a VM with

  • 16 CPU cores
  • 112 GB RAM
  • 32 TB LRS SSD data disk with 6.25 TB/s throughput (160k IOPS)
  • 127 GB system/boot disk
  • 800 GB SSD local (non-persistent) temporary storage

To put that in perspective relative to the cost of other Azure storage options:


Block Blobs cannot be attached to VMs as disks of course, only Page Blobs can. Also, the D14 size VM is the only one that can have 32 disks currently.

The local SSD temporary non-persistent storage on the D14 VM provides 768 MB/s read throughput


or 384 MB/s write throughput




2 responses

  1. Pingback: Options for using a Veeam Backup Repository on an Azure Virtual Machine | Sam's Corner

  2. FYI: on top of the disk throttling there is also throttling on a per-VM basis. Meaning you do not get the full 5000 x [# of disks]… each VM in the portal specifies its own limit. eg. the DS14 taps out at 50,000 IOPS regardless of how many disks you have attached. It would seem that the throttling for premium storage is based both on throughput (MB/s) and # of IOPS both for the disk and VM. eg. no one disk can exceed 5000 IOPS and no combination of disks can exceed XXXX IOPS (depends on VM)

    February 4, 2015 at 6:08 am

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.