Veeam vPower Instant Recovery

I was talking with some colleagues the other day and we were going through the different ways that one of our customers could avoid the purchase of a SAN, but still have very good RPO and RTO. We have recommended a SAN several times, but local storage has always “worked” for them and they haven’t felt the pain of a VMware server with local storage and several VM’s go down. We discussed lower cost options such as the HP P2000 G3 and reusing existing 2.5″ SFF drives, or maybe an EMC VNXe series. But Ben was still skeptical and felt that the customer would not be interested.

Then I remembered the new vPower Instant Restore feature with Veeam. Basically, with this feature, Veeam presents an NFS datastore to an ESX(i) server that you select and then does some magic to the backup files and presents the virtual machine in native VMware format back to the VMware server. Then it registers the VM into inventory and will even power it on.

The really cool part is that you do not have to wait for the VM to be restored back onto local storage or your SAN… it actually reads and writes to the hard drives in your backup server. Then once you are back up and running you have two choices:

  1. finish out the day and shut down the VM, then move the VM back to your normal datastore (be it a SAN or local storage)
  2. use Storage vMotion to move the VM to your production datastore while it is running. This requires at least Enterprise level VMware licensing.

The Process

To show this feature I took some screenshots of my cluster, I deleted NS2 which is my production DNS server for my blog and other sites. Then restored it using vPower and the Instant Recovery feature. Below are the steps and the screenshots. (As with all pictures on my blog if you click them they will open the full version for easier viewing.)

Step 1. Open Veeam and navigate to the “Backups” section and right-click the VM we want and select “Instant Recovery”

Step 2.Select the restore point you want to restore.

Step 3.Select a Host, a VM Name, and a resource pool to restore to. Also, check if you want it powered on and connected to the network automatically. (DO NOT CHECK THESE IF YOU ARE JUST TESTING, also if just testing pick a name that is not already in use)

Step 4. Choose if you want to write changes to a different datastore. If your backup server has a slow link, or a limited number of IOPs this may be a good choice.

Step 5. Give a reason for the recovery (if needed)

Step 6. Review the settings and click Next.

Step 7. Wait until Veeam says its ready. At this point, the temporary NFS datastore has been mounted to the server and the VM has been added to inventory and powered on if you told it to. You can click “Finish”.

At this point your VM will be powered on and working as if nothing happened. The only step left is to Storage vMotion (or shutdown and move after-hours) the VM to its normal data store.

After you move the VM to a normal datastore you will need to un-publish the NFS datastore. Basically, this will tell Veeam that it is ok to stop the NFS magic to the Backup file. To do this go into the Veeam console and select the “Instant Recovery” item on the left and right click on the item to un-publish and click it. Then a dialog box will show you the progress and eventually look something like this when it is completed.

It should be noted that this entire process took about 10 minutes. A normal restore operation could take much longer because the data must move across the network back to the SAN, so if your restoring 1TB that will take several hours. If you use Instant Recovery, performance will be degraded but that VM will be back online in a much shorter amount of time, and then after hours you can move the VM back to production storage and be ready for the next day without almost any downtime.

Clearly, this is not a replacement for a SAN, it will not provide the uptime that VMware HA or FT will provide, but if you absolutely cannot afford a SAN and prefer local storage for some weird reason I would definitely recommend Veeam because of its Instant Recovery Features.



Share This Post

5 Responses to "Veeam vPower Instant Recovery"

  1. “they haven’t felt the pain of a VMware server with local storage and several VM’s go down”

    Justin, you turned me on to the P2000 and I was ready to sign the cheque. But now I’m having doubts. Why is a Proliant server more likely to go down than a P2000 – they’re essentially the same bit of kit aren’t they? A P2000 doesn’t prevent me from feeling the pain of ALL my VMs going down when the SAN fails.

    I’ve written about my doubts here:

    I don’t know if you do requests, but I’d love a blog by you explaining why a SAN beats Local Storage for a SMB.

  2. “I don’t know if you do requests, but I’d love a blog by you explaining why a SAN beats Local Storage for a SMB.”

    It doesn’t. Single box = local storage is just fine and doesn’t make sense for a san. You have a single point of failure period.

  3. SAN units tend to have rather robust redundancy; it’s not impossible to lose a SAN in the bat of an eye, but it’s far less likely than losing a regular server. Horses for courses, and most small to medium companies probably don’t really need that level of performance or reliability.

    Unfortunately, we aren’t in such a position, so we have several.

  4. Grenage,

    I agree, a SAN is typically going to tell you it is sick and give you a chance to repair the problem before just falling over dead like a typical server can… How many people have had a raid controller fail in a dell ? In a san that wouldnt take you down, it would just fail over to the other controller, in a DAS environment that server would be dead until the controller was replaced.

Post Comment