I need to move data from three data disks on an active production Azure VM to a newly deployed utility server. The source VM must stay online throughout the migration, and this is a one-time move. What approach would you recommend for the initial copy and final synchronization, and are snapshots, Azure Site Recovery, or tools such as Robocopy suitable?
4 Answers
The right choice depends on the consistency requirement. Snapshots are convenient for a crash-consistent point-in-time copy, while Robocopy or another in-guest tool is better when you need to capture changes up to the cutover. Also check the destination disk performance tier, since Azure storage throughput can be much lower than expected unless the disks and VM are sized appropriately.
Azure Site Recovery is another option when you want ongoing replication and a controlled cutover, including replication to another availability zone in the same region. For a simple one-time file migration it may be more setup than necessary, but it is worth considering if you need minimal downtime or a stronger recovery workflow.
If you need a point-in-time copy rather than continuous file-level synchronization, snapshot each source data disk, create new managed disks from those snapshots, and attach them to the pre-staged destination VM. This is quick and can work across subscriptions or regions, but the result is generally crash-consistent unless the workload is quiesced first.
For a one-time migration, an in-guest copy is usually straightforward. Use Robocopy for an initial seed while the production VM stays online, then run another pass to copy changes and finish with a short cutover window. Make sure the applications are quiesced or stopped for the final sync if you need application-consistent data.
That fits the situation well. The destination is a new utility server, so doing the initial copy followed by a final differential pass seems practical.