Blog Post

Azure Storage Blog
3 MIN READ

Protect Azure workloads with VM level consistency using Agentless Crash-Consistent Restore Points!

aarthiv's avatar
aarthiv
Icon for Microsoft rankMicrosoft
Jul 18, 2023

Today we are happy to announce public preview support for multi disk crash consistency mode in Virtual Machine (VM) restore points. A crash consistent VM restore point is an agentless solution that stores the VM configuration and point-in-time write-order consistent snapshots for all managed disks attached to a VM. This is same as the status of data in the VM after a power outage or a crash.

 

VM restore points, announced in July’22, enabled reliable restoration of disks and VMs for data loss, corruption, disaster recovery, and infrastructure maintenance incidents. Using VM restore points, Azure Backup and ISV partners such as Commvault and Veritas offer BCDR solutions for customers. VM restore points are incremental, where the first VM restore point stores a full copy of your data. For each successive restore point of the VM, an incremental copy i.e., only the changes to your disks are stored. The incremental design enables you to benefit from the data protection of frequent backups while minimizing storage costs. You can also use the built-in copy functionality to copy your VM restore points to any region of your choice to get protection from regional failures.

 

 

Key Benefits of Crash Consistent Restore Points

  1. Agentless solution
    • Using agents for VM restore points is considered as a security, compliance, and management overhead by some partners/customers.
    • Crash consistent restore points directly takes the multi-disk consistent snapshots from the host machine thereby removing the overhead of an agent inside the VM.
  2. OS agnostic support
    • As an agentless solution, there is no dependency on the guest operating system (OS). All Windows and Linux OS types are supported by crash consistent restore points.
    • Erstwhile unsupported Linux OS versions, 32-bit OS systems, Windows VMs with ARM64 etc. with application consistency mode are now supported with crash consistency mode.
  3. High Frequency support
    • Crash consistent restore points support 1 hour frequency enabling lower RPO for applications running on Azure VMs.
  4. VM Level Consistency
    • Prior to VM restore points, customers/partners had to use managed disk snapshots which are taken at per disk level. Due to this consistency at VM level could not be guaranteed and it was also cumbersome to manage.

Resiliency solutions with crash consistent VM restore points

Azure Backup: Providing first class backup support using VM restore points

Azure Virtual Machine Backup enables you to create an enhanced policy to take multiple snapshots a day. This allows you to protect your virtual machines with Recovery Point Objective (RPO) as low as 4 hours. Azure Backup now supports crash consistent restore points (in private preview). Please enroll here to use the capability.

 

Azure Backup will enable customers to protect a wider set of Virtual Machines (VM) running Linux distributions that are not on the current support matrix as well as VMs that do not use Azure extensions using crash consistent restore points.

- Aravindan Gopalakrishnan, PRINCIPAL PDM MANAGER – Microsoft, Azure Backup

 

Zerto – An HPE Company: Delivering Consistency efficiently with crash-consistent VM restore points

Zerto an enterprise-class business continuity and disaster recovery company, is one of the first ISV partners to integrate the new crash-consistent snapshot capability into their product. This integration will enable whole VM protection with crash-consistent snapshots across multiple volumes.

 

“Multi-volume consistency protection is one of the most sought after features by Azure Customers”.

- Shannon Snowden, Senior Product Manager - Zerto, an HPE Company

 

“Multi-volume virtual machine (VM) level crash consistency is critical in disaster recovery protection. With the new Azure crash consistent snapshot capability, it enables Zerto to create VM level crash-consistent restore points using underlying snapshots.”

- Sandra Biton, Engineering Group Manager – Zerto, an HPE company

 

Zerto 10 introduces multidisk consistency for Azure VMs, which protects VMs to, from, and within Azure with complete disk consistency. Moving away from snapshot-based replication, multidisk consistency for Azure VMs now leverages a new restore point API, offering an easier and more efficient way to manage replication and recovery operations.

 

Get Started

  • Click here to enrol in public preview of crash consistent VM restore points,
  • Learn more about VM restore points.

Please share your feedback or questions in the comments section below.

Updated Jul 18, 2023
Version 1.0
  • EitanBlumin's avatar
    EitanBlumin
    Brass Contributor

    That sounds interesting. I'd love to learn more about how that works.

  • Hi EitanBlumin , creating crash-consistent restore points is hassle-free, as it eliminates the requirement for VM extensions like VSS writer, unlike the application consistency mode. There should not be any issue in creating crash consistent backups for SQL server machines or any databases for that matter. Please feel free to reach out if you have any further inquiries or need more information.

  • EitanBlumin's avatar
    EitanBlumin
    Brass Contributor

    I'm not familiar with SRDF from EMC.

    But I am familiar with application-consistent backup solutions that utilize SQL Server VSS to perform backups.

    Whether SQL Server will be able to return from such a backup depends on how this backup is run behind the scenes. Specifically, in what order it's backing up the storage disks.

  • HeatfanJohnM's avatar
    HeatfanJohnM
    Copper Contributor

    Why is this incompatible with SQL Server machines?

     

    Is SQL Server unable to return to a consistent state after a crash or power failure?

     

    How is this different from using Asynchronous SRDF from EMC with a SQL Server database?

  • EitanBlumin's avatar
    EitanBlumin
    Brass Contributor

    Someone should mention that this is incompatible with and very dangerous for SQL Server machines.