yyaazz's tech thoughts

In the previous part I started to present how NetApp Virtual Storage Console and SnapManager for Virtual Infrastructure integration works. I’ve created a backup then checked it on the controller and finally made a restore. In that example the restore handled the whole volume, so the other VMs running there were also affected (restored). In this post I’ll show how can be the restore handled more efficiently and how the single file restore works.

For today’s example I created a backup called ‘smvi-singlefile-2′ of the same virtual machine, then removed the kernel of the operating system. With the single file restore I can mount the backup today and copy out whatever I need. Let’s see how to do it:

The first step of the wizard lets me choose some basic options, the source and destination VMs, notification and the mount expiration.

In the next step (Guest File Restore Resource Access Type) I have to choose between ‘Self-Service’ and ‘Limited Self-Service’.

Self-Service: The Backup and Recovery administrator creates a restore session using Backup and Recovery capability. Users can then install Restore Agent on the destination virtual machine, browse the mounted backups on a guest virtual machine, and recover the individual disk file.

Limited Self-Service: The Backup and Recovery administrator finds the backup copy within a user- specified range of backups and attaches the backed-up disks to the destination virtual machine. The user can then run Restore Agent on a destination virtual machine, browse the mounted backups, and recover the individual disk file.

The Limited Self-Service is comfortable for me now as I just want to copy out the file from the backup (no Restore Agent in this example).

I can browse the backups of the VM now to choose for restore. The wizard has one more summary page where the process can be started. Let’s see how magic happens:

VSC exports the snapshot as a NAS datastore, and attach to the ESXi host which runs the VM.

The destination VM got the appropriate VMDK as a new Hard Disk (Hard disk 2 in my example). At this point I have to do the restore in the VM:

• list the current disks (fdisk -l)
• rescan the scsi bus if necessary (via the sysfs)
• list the current disk again and check the new one

• mount the partition

• copy the file(s)
• umount the partition

While in the previous post I restored the whole datastore with five VMs, in this example only one file in only one VM. On the other hand this procedure were online while last time the running VM restarted.

The 3rd part will show how to create a SnapMirror to replicate the backup to a remote site and how to restore from that.