SQL Cluster Restore to Azure

[harryz123]

Hi All,

I have a project to do a test restore of an on prem SQL FCI Cluster (2016 SQL & Windows 2016) host in a vSphere environment. We backup the cluster using Veeam Agent for Windows. From what I understand, the restore would convert the disks to local disks to the VM once restored in Azure. I assume that once this happens, the cluster would no longer work. How would I go about ensuring SQL would work? The solution could have something to do with forcing the cluster in some fashion or breaking the cluster. It is only a test and my main concern is to get it working again without recreating anything if possible.

[Andreas Neufert]

I guess the way to go is to restore the database and not the servers.

[johan.h]

I'd imagine that instead of running a SQL Server FCI cluster of Azure VMs you would want to run a Azure SQL PaaS offering (Azure SQL Database or Azure SQL Managed Instance). Is there a reason you don't want to use an Azure managed version of SQL?

This would take some work to make happen, and perhaps some specific features you are relying on prevent this? What is the desired state you're looking for? If this is only a test, is the end goal simple disaster recovery? If so perhaps a standalone SQL server will do?

The import and export process of moving data from your on-premises SQL server should be possible using native tools. You could, in broad strokes, use Veeam to temporarily publish a backup (attach a backed up SQL database) to a server in your environment: https://helpcenter.veeam.com/docs/backu ... ml?ver=100. And then, for example, use SQL Management Studio / SqlPackage to complete the export process, after which you would import this into Azure: https://docs.microsoft.com/en-us/azure/ ... powershell

If you are planning to run a SQL FCI cluster on Azure VMs, what form of shared storage are you planning to use? Are you doing this using Azure Shared Disks, Storage Spaces Direct or a Premium File Share? Or would you be thinking of using SQL AG? From a cost perspective all of these would have to be evaluated. But for restore you may still use the Veeam publish functionality on-premises, and then use native SQL tools for the actual migration.

That said, I'm not sure exactly what you want to achieve, feel free to DM me in follow-up.

[harryz123]

I am not planning on restoring the cluster. All management wants is one of the two nodes in the cluster operational in Azure (does not need to be in a cluster role). They know we cannot restore the cluster in working condition so they want to test restoring just one of the two nodes. That is my challenge. How do I get this node functioning as easy as possible? Do I just create a new VM with SQL on it and then restore the data to the new disks in Azure? Can I do a complete restore and then force the cluster somehow?

[johan.h]

I would not recommend a full restore of a cluster node. You're restoring to dissimilar hardware anyways. I would focus on database restore only.

(You could see what happens if you restore a VAW backup of a single node, and hope it works of course).
https://helpcenter.veeam.com/docs/backu ... ml?ver=100

[harryz123]

So would creating a new VM with SQL on it and restoring the individual volumes be ok? Then I'd have the SQL DB restore the DB's on those volumes/disks himself in SQL.

[Dima P.]

Volume level recovery should help in such case, additionally you can restore specific databases via file level recovery to a VM .