Backup and restore: Deploying the VMware VDP Appliance
Before you can deploy the vSphere Data Protection (VDP) appliance, you need to know its capabilities and its limitations. In another blog post on installing and configuring VDP we followed the steps needed to prepare for the setup, deploy the appliance from a template and do the initial configuration.
Once you install VDP, the next steps are to:
- use the VDP appliance to back up some VMs
- restore a VM to a previous state
- restore it completely if the original got lost
As discussed earlier, once you set up your appliance you will notice a new VDP item on the left of your vSphere Web Client interface. From there we need to do all administration tasks including backups and restores.
You need to connect to your VDP of choice because you may have more than one to account for the capacity limitations or other administrative reasons.
After you connect, you will see the getting started tab, which can be used to quickly create a backup job or restore a VM. Although those tasks and much more can be accessed from their own tabs, I have created my 1st backup job from this screen.
The wizard starts with the important choice of VMs to be backed up as part of this job. You can choose entire data centers with all of their VMs, or drill down to the level of hosts and clusters. I wish I had the option to choose folders of VMs as I wanted to create a task to back up some infrastructure VMs, like domain controllers, firewalls and antivirus management server (I already have them in a folder).
It’s also critical to decide how often to perform this backup. Notice you cannot specify the time of backup here because it is controlled by the backup window. The backup window is configurable for the appliance and not on the level of each job.
In some environments you may need to keep your backups forever or at least for a long time. For some jobs a retention period of 60 days is enough.
Although there is no option to create a one-off job or a yearly job, you can do this by creating a regular job and disabling it after it runs for the first time.
As with everything in IT, always pick a representative name for your backup job. The following screen will show a summary of our choices so far.
Review and hit finish, then wait for several minutes for your 1st backup job to be created.
The Backup Tab
If we wait long enough the backup job we just created will be initiated according to schedule, but since we are eager to try it, we can manually start it in the backup tab.
Here we can also enable/disable, delete, edit or create new jobs.
Everything that our backup job performed gets logged as tasks performed on objects in our vSphere infrastructure. Those events can be monitored on different levels like individual VMs, hosts, data centers etc.
Creating a user for each purpose helps us understand what is happening and why, as those tasks are all performed by the vdp.user we have created for the VDP.
Restore a VM to its previous state
It seems that an admin has messed up the configuration of one of our Vyatta Linux-based virtual routers and he had deleted it in frustration (or to cover his tracks).
No worry. We can restore it back to what it was a few days ago even though it is deleted. If it was not totally lost and only its configuration messed up, we will have to go through the same wizard, but with a slight change in the choices.
The only big difference is that the restore will take much less time. Only changes will be restored because most of the data between the two points in time is identical.
Pick a restore point to recover the missing (or the messed up) VM.
Choose what you want to call it, and the destination where you want to restore it. If a VM is completely missing, then maybe you want to give it the same name that it used to have. If you want to recover another copy of the VM, you can leave the date that was annexed to the name by the VDP, or chose a different name.
Notice that the option “Restore to the Original Location” is grayed out. This option will be available if we are restoring a VM that is still there and not completely missing.
Also notice that since we are recreating the VM, a new VM will be created. If we are restoring a VM that was not deleted to its previous state, we will be replacing one VM and not creating any new VMs.
It took about 2 minutes to restore this 4GB VM across hosts and from one SAN to another. Depending on your environment and size of your VMs, the time may vary.
The most important thing in the reports tab is ability to see how much capacity of this VDP is already in use. The reporting is also used to see any status of backup jobs performed on each virtual machine, and an ideal place to reach the events and tasks related to VDP operations.
In addition to configuring email alerts, the last tab is the place to configure your backup, maintenance and blackout windows.
So what’s the difference between the blackout window and the maintenance window? According to the VDP FAQ:
“The maintenance window is the portion of each day reserved for performing routine server maintenance activities such as integrity check validation. Whereas the blackout window performs server maintenance, which requires unrestricted access. Server maintenance such as garbage collection that deletes the orphaned chunks of data that are no longer referenced within any backups stored on the system.”
The practical difference is that backups can be initiated during the maintenance window (although doing so affects both the backup and maintenance activities). However, Backups and administrative activities are not allowed during the blackout window, though restores can be performed.
If you have more than one VDP, you may want to adjust those windows so backup jobs performed on the same VM by different VDPs do not intersect and cause each other to fail.
Wow! Finally I have a great and free backup solution for my virtual infrastructure, and I can sleep well at night knowing that my VMs are safe. One important hint to keep in mind: do not store the VDP appliance on the same storage backend as your VMs because you need to have backups on an extra storage for when the main one fails.
It depends on your budget, but even if you use a cheap one that is much slower than your production storage, you will be much safer than leaving the chance of losing them both at the same time. I got myself a Synology NAS and put it in a remote and safe corner of the office just for this purpose.