What’s the Best Way to Migrate Azure VM Data Disks Without Downtime?

For a one-time migration, an initial copy followed by a final delta sync is probably the most practical option. Robocopy can seed the destination while the production VM stays online, then run again near cutover to copy only changed files. You can briefly pause applications or writes for the final pass if you need a clean handoff.

Azure Site Recovery is worth considering if you want continuous replication and a more controlled cutover. It can replicate the VM to another availability zone in the same region, depending on the configuration, although it may be more tooling than necessary for a single file-data migration.

If the disks can be copied as a point-in-time state, snapshot all three managed disks and create new disks from those snapshots, then attach them to the pre-staged utility server. This is straightforward and can also work across subscriptions or regions, but the snapshots are crash-consistent rather than guaranteed application-consistent.

Azure File Sync is another cloud-native design: sync the source VM to an Azure storage account and then from storage to the new server. It avoids copying directly between the machines, but for a one-time move, a direct seeded copy plus final synchronization may be simpler and faster.

Whichever method you choose, validate the destination’s disk type and IOPS beforehand. Azure storage performance can be noticeably lower than on-premises systems unless the disks and VM size are provisioned appropriately.