Share via

Best Practice for Azure DR using ASR When DR Region Should Not Host Any Pre-Provisioned VMs

Paul 20 Reputation points
2025-12-04T02:21:48.5766667+00:00

Hello Team,

I am designing a Disaster Recovery (DR) strategy for an Azure environment and would like Microsoft’s best-practice guidance.

Scenario:

Primary region hosts ~50 Azure IaaS VMs (Web, App, DB, AD, middleware).

SQL is running on Azure IaaS VMs (SQL Server Always On is being evaluated).

DR region must be cost-optimized.

Customer does not want any VMs pre-provisioned in the DR region.

Only network components (VNet, subnets, NSGs, UDRs, DNS zones, and Recovery Services Vault) can exist in DR.

Requirement is to perform one-click failover during a regional outage using Azure Site Recovery (ASR).

During failover, all 50 VMs must be automatically created in the DR region, brought online in the correct sequence, and application made accessible.

No manual creation of VMs at the time of failover is acceptable.

Questions:

Is ASR Recovery Plan the recommended approach for orchestrating end-to-end failover (VM creation, sequencing, post-failover automation, LB/DNS updates) in a scenario where no VMs exist in the DR region before failover?

Does Microsoft recommend pre-creating certain components in the DR region (such as Load Balancers, SQL replicas, or basic infrastructure VMs like DCs), or is it fully supported to have zero compute in the DR region and let ASR create all VMs automatically?

For SQL on Azure VMs:

Should we rely on ASR replication alone, or is SQL Always On Availability Groups recommended for better RPO/RTO?

  If SQL AG is used, how should automated AG failover be integrated into ASR Recovery Plans?
  
  **For automation:**
  
     Are Azure Automation Runbooks or Azure Functions the recommended way to perform tasks during failover (DNS updates, reconfiguring backend pools, application warm-up)?
     
        Is there any official Microsoft sample repository for ASR Recovery Plan automation scripts?
        
        **For large environments (~50 VMs):**
        
           Are there any documented limits or performance considerations when performing a bulk failover where ASR needs to create many VMs at once in the DR region?

Goal:
To implement a Microsoft-aligned DR solution where:

DR region stays cost-optimized (no running VMs)

ASR performs full automated failover (one click)

SQL DR is handled as per best practice

Orchestration and automation follow Microsoft recommendations

Any official guidance, reference architecture, documentation links, or best practice patterns would be greatly appreciated.

Thank you.

Azure Site Recovery
Azure Site Recovery
An Azure native disaster recovery service. Previously known as Microsoft Azure Hyper-V Recovery Manager.

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Q&A Assist
    2025-12-04T02:22:04.2533333+00:00
    AI generated content. This question contains an answer created with AI from Q&A Assist. Learn more

    For your disaster recovery (DR) strategy using Azure Site Recovery (ASR) in a scenario where no virtual machines (VMs) are pre-provisioned in the DR region, here are some best practices and recommendations:

    1. ASR Recovery Plan: Yes, using an ASR Recovery Plan is the recommended approach for orchestrating the end-to-end failover process. This includes the automatic creation of VMs in the DR region, sequencing their startup, and executing post-failover tasks such as load balancer and DNS updates. The Recovery Plan allows you to define the order in which VMs are brought online, ensuring that dependencies are respected.
    2. Pre-Creation of Components: Microsoft supports having zero compute resources in the DR region. ASR can create all VMs automatically during a failover. However, consider pre-creating essential network components (like VNets, subnets, and NSGs) and potentially critical infrastructure components (like domain controllers) to ensure a smoother recovery process. Load balancers and SQL replicas can also be pre-configured if needed, but it is not mandatory.
    3. SQL on Azure VMs: While ASR replication can be relied upon for DR, implementing SQL Server Always On Availability Groups (AG) is recommended for improved recovery point objectives (RPO) and recovery time objectives (RTO). If you choose to use SQL AG, you can integrate automated failover into your ASR Recovery Plans by scripting the failover process within the Recovery Plan's automation tasks.
    4. Automation: Azure Automation Runbooks or Azure Functions are indeed recommended for performing tasks during failover, such as DNS updates and application warm-up. These tools can help automate the necessary configurations to ensure a seamless transition.
    5. Official Samples: While the context does not specify a repository for ASR Recovery Plan automation scripts, you can typically find official Microsoft documentation and sample scripts through the Azure documentation portal.
    6. Performance Considerations: For large environments with around 50 VMs, ASR is designed to handle bulk failover scenarios. However, it is advisable to review any documented limits or performance considerations specific to your deployment and test the failover process to ensure that it meets your expectations under load.

    By following these guidelines, you can implement a cost-optimized DR solution that aligns with Microsoft’s best practices, ensuring that your ASR setup is effective and efficient.

    References:

    0 comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.