Skip to main content

Hello,

We have determined with the help of CV support and Azure support that the fastest possible restore of an entire VM within the Azure environment (even within the same region) is 60Mb/s. The limitation, according to Azure, is that CV writes the VM as a single page blob and the max throughput for that is 60Mb/s.

At this speed, restoring an entire VM of any significant size would take far too long to be acceptable in an emergency situation.

Has anyone run up against this limitation and how have you worked around it? Or does anyone have any suggestions on how we might do so?

Thanks in advance for your suggestions!

@Neil Winking , I’m going to convert this to a conversation to encourage more members to contribute their experiences :nerd:

Hi @Neil Winking,

 

What Maintenance Release version are you on here? - In V11 FR26 Maintenance Release 11.26.7 contains two HotFixes specifically for this:

Azure full VM restore may extend for a longer duration with low write throughput speed.

 2162

Azure VM restore may be slow.

 2151

 

Also, are you restoring Managed Disks? 

 

Best Regards.

Michael

Hi @MichaelCapon,

Thanks for the info. We are on LTS V11.24.23 currently. I’m wondering if these HotFixes address a slowness specific to 11.26 or if we would see improved performance from an upgrade.

Yes, we are restoring managed disks.

Thanks again,

Neil

Hello Neil,

what I did in the past to get around slow restore speeds and recover very large Azure machines quickly was to leverage Commvault IntelliSnap functionality for Microsoft Azure:

Azure IntelliSnap

I created snap_primary copy, set the retention to have e.g. 7 most recent snapshots kept (depending on your requirements), then used the backup copy operations to copy data to Azure storage for longer retentions.

Advantage of this solution was the ability to recover whole VMs in about 10 minutes, no matter the size (I used it for 24TB machines), but only for most recent recovery points. However, usually when we need to restore whole VM due to crash, it’s the most recent copy that we use to restore from.

Best Regards,

Marcin

Hi @Marcin Dziewanowski,

That’s a good idea. I’m going to look into that and discuss with the team. I think it would meet our needs in most scenarios.

Thank you,

Neil

Terms & ConditionsCookie settings