Technical Thursday – Restore VM from Recovery Services vaults (concept)

There is a sensitive topic which is important for business and IT as well. This is the BCM (Business Continuity Management). And its technical part the DRP (Disaster Recovery Plan) which related for us now. This is the topic which could cause some sleepless nights for us. Then you hear the cloud solve your and your company’s problem…so you decide to backup your on-premises and Azure VMs with a cloud solution especially Recovery Services vaults.

To get an overview you should start to read the following page: Azure Backup Documentation This is a very usefull and detailed documentation and contains the most important points such as Concepts, Samples, references. Regarding VMs you will find the following pages:

• Back up Azure virtual machines to a Recovery Services vault
• Use the Azure portal to restore virtual machines

The backup is quite easy because you create Recovery Services vaults for your VMs then you configure the schedule of backup then retention time and it works independently you have unmanaged or managed disks under your VM. 😉

Restore is our fear because the consistency and the availability is critical in a disaster situation. Additionally we have to be ready to restore the unmanaged and managed disk based VMs as well.

Restore unmanaged disk based VM

This situation is that when you have VHD files on a storage account which is used by your VM as OS disk and/or Data disks. Although a simple scenario is written in Use the Azure portal to restore virtual machines article you should be ready on conceptual level to restore a VM which is a little bit complex (example: it uses multiple NICs or it was created from a labyrinthine ARM Template). In this case you cannot restore the whole VM from Recovery Services vaults. Your only one option is to restore the VM related VHD files to a storage account then recreate the VM.

The concept is here for this scenario:

1. Stop (deallocated) the target VM
2. Delete target VM object (ONLY the virtual machine object)
3. Delete vhd files which belong to target VM
4. Choose the right restorepoint from Recovery Services vaults which belongs to target VM
5. Restore OSDisk and DataDisks to dedicated storage account.
6. Move restored vhd files to original location
7. Re-create target VM with ARM Template

I will provide some materials for this  scenario on next week.

Restore managed disk based VM

Here you do not have VHD files under VM but you have very good managed disks. These disks are standalone objects on Azure level so you can manage them on that way. Nevertheless when you restore the disks from Recovery Services vault you will realize that you will have one vhd files for each restored disks which belong to your restored VM. Why? The answer is in the article you can find above:

Templates are added as part of restore disks for recovery points taken after March 1, 2017. They’re applicable for nonmanaged disk VMs. Support for managed disk VMs is coming in upcoming releases.

Accordingly we have to wait some days or weeks until Microsoft fully supports managed disks related restore. No problem, we can apply some workarounds until that. 🙂

Managed disk restore steps (there is one additional steps here):

1. Stop (deallocated) the target VM
2. Delete target VM object (ONLY the virtual machine object)
3. Delete managed disks which belong to target VM
4. Choose the right restorepoint from Recovery Services vaults which belongs to target VM
5. Restore OSDisk and DataDisks to dedicated storage account. (Result will be VHDs)
6. Convert VHDs to Managed disk
7. Move restored managed disks to original location
8. Re-create target VM with ARM Template

I will provide some materials for this  scenario on next week.

To be continued on next Thursday…