Migrating Windows Failover Cluster to new VMware Environment

[holtadams]

Veeam Product Team, this is related to a Veeam Support Ticket (Case #04865757)

I'm using Veeam Backup and Replication server to migrate a VMware environment to IBM Cloud. I've previously used Veeam with Backup Proxies at both on-premies and remote sites to successful migration VMs (creating backups and then replicating into VMware as a 2 step migration process), but for my current migration I have a Windows Failover Cluster that is presenting a few new challenges. The nodes of the Failover Cluster utilized physical RDMs for shared storage, thus I've deployed Veeam Agents on these nodes and have successfully backed up the nodes and all of their drives (share and non-shared drives) with a Veeam Backup Failover Cluster job. However when I use Instant Recover/Migrate or a Bare Metal server recovery (using Recovery Media), the shared drives are not provisioned.

I've exported the shared drives from the Veeam backups to create VMDK files, and manually added additional Hard Drives to the VM Nodes (with existing drive options) to attach the VMDK files. When the VMs are powered up, I can see the Disks using Disk Management, but when I try to bring the Disks online, I get a error message stating the drives are "reserved" and that Failover Cluster Manager should be used. When I start the Failover Cluser Manager, there are no clusters shown as being defined.

I'm planning on removing the Harddrives that I created with the VMDK files, and adding Harddrives defined as RDMs, hoping the recovery process makes the drives accessible.

Has anyone else successfully migrated a Windows Failover Cluster with Veeam? Any best practices that can be shared would be greatly appreciated.

[HannesK]

Hello,
and welcome to the forums.

As we are talking about physical machines (a RDM is physical), only methods for physical machines will work.

When I tested IBM cloud last time for CDP, there was only VSAN available as datastores. From the documentation I see that RDMs are not available for VSAN and as far as I see the method you used cannot work https://docs.vmware.com/en/VMware-vSphe ... 1BCFE.html

Shared VMDKs should work: https://kb.vmware.com/s/article/74786

If you restore the nodes (bare metal restore) and then do a volume restore of the shared disk, that might work.

In general I would recommend migrating to technology that is virtualization friendly (meaning always-on availability groups). Because shared VMDKs have downsides like "no snapshots" (see KB article above). Support also already pointed out that you could deploy block storage (iSCSI LUN, RDM), but not sure whether that is possible.

Best regards,
Hannes