Backup Process

The backup operation includes the following stages:

  1. Discover virtual machine in the Azure cloud based on subclient settings.
  2. Create a client computer entry for each virtual machine discovered.
  3. Identify the virtual machines targeted for backup and the VSA proxies that can be used to perform backups. VSA proxies can be installed on Azure virtual machines or on local physical or virtual machines.

    For Azure Classic backups, the VSA proxy must have an Azure management certificate installed.

  4. Identify the disks associated with virtual machines targeted for backup.
  5. Assign backup streams to proxies in round-robin sequence.
  6. Use the Virtual Server Agent (VSA) to communicate with Azure to download data.
  7. Create virtual machine snapshots in the Azure cloud. By default, Azure backups are crash consistent backups. There are two methods of backup :
    1. Cyclic Redundancy Check (CRC): Read extents from snapshots, compute CRC and write to backup media. After backup completes, delete the snapshots.
    2. Changed Block Tracking (CBT): With this option enabled by default at the subclient level, only the extents that have changed in between consecutive CBT-enabled jobs are backed up. This is completed with Azure API comparing the modified extents between the two consecutive jobs, reading these extents and then writing it to backup media. Once the backup is complete, upload the details of current snapshot to the index, delete the snapshots of the previous CBT-enabled job that was used for comparison and save the current job’s snapshots to be used in the next CBT incremental job for validation.

      NOTE: Microsoft has a limitation on bandwidth for the snapshots created on unmanaged disks that reside in a premium storage account. As a workaround, a staging blob is created from the snapshot. Then, the backup operations read from the staging blob. After the backup job completes, the staging blob is deleted.

  8. Clean up the Azure mount point, and complete the backup job.  

Last modified: 9/10/2019 8:47:30 PM