Large Server and Offsite Backup Copy

[bg.ranken]

We currently have a large server that we are backing that we need to set up an offsite backup copy job for. The settings seem right to me and I've done backup copy jobs before, but not with a server of this size.

Here are my current specifications:

• Server size: 4TB, 3TB Veeam backup file size
• Backup Mode: Reverse incremental
• Restore points to keep on disk: 90
• Create active full backups periodically: monthly every three months
• Average daily incremental size: 5GB
• Average incremental job run time: 1-2 hours
• Backup copy target: Offsite Windows server with a QNAP NAS device attached via iSCSI
• Link between sites: 20MBs

What I am planning on setting up is a backup copy job with the following settings:

• Copy every: 1 day
• Restore points to keep: 90
• GFS: Weekly-0, Monthly-0, Quarterly-4, Yearly-1

I'm concerned that the backup copy job may not have enough time in the daily intervals to make a full backup checkpoint for the GFS policy. I'm also worried that the active full backups done quarterly will copy in full to the offsite location rather than just an incremental portion. With our speed it may cause the backup copy job to run indefinitely for a few weeks after every active full since our WAN speed is not that great.

Should I turn off my periodical active full backups? We only have them set because we don't currently have SureBackup configured and I want to make sure the backups are good.
Should I increase the copy interval of the backup copy job to ensure the agent on the remote site can finish a full GFS checkpoint?
Should I set the copy interval of the backup copy job to start before the daily backup or just after it normally finishes?
Are there any other configuration setting changes I might want to make?

[foggy]

Randall, you should seed the initial backup copy job run to avoid transferring the entire VM data, only changes will be transferred afterwards regardless of periodic active fulls on the original job. Also, to create GFS restore points, Veeam B&R does not transfer the entire VM data, but rather uses existing full backup from the regular chain.

[bg.ranken]

Foggy, we plan to put the NAS at the main site and get the backup copy running to it and then ship the NAS offsite to continue the backup copy job. I'm glad to hear that the active full backups won't transfer which is nice.

In regards to the GFS, from what I've observed the GFS does leave the backup the existing full backup file, however to continue the current chain it seems to make a new full backup from the existing full and the oldest incremental. This would mean that when the retention reaches the GFS point it would have to copy 3TB of information, and that's what I'm worried will take more than 24 hours in our environment, especially seeing as it's copying and writing the file to the same NAS. Is this a correct way of looking at it?

Also, can you answer the questions about scheduling as well?

[foggy]

This would mean that when the retention reaches the GFS point it would have to copy 3TB of information, and that's what I'm worried will take more than 24 hours in our environment, especially seeing as it's copying and writing the file to the same NAS. Is this a correct way of looking at it?

Yes, and this solely depends on the NAS I/O capabilities.

Should I increase the copy interval of the backup copy job to ensure the agent on the remote site can finish a full GFS checkpoint?

Yes, you can increase the copy interval, if having less frequent restore points is acceptable.

Should I set the copy interval of the backup copy job to start before the daily backup or just after it normally finishes?

It is typically recommended to start both jobs at the same time so that backup copy sit and wait for the new restore point to become available.

[bg.ranken]

Thanks for the response foggy. I have a few more questions:

Yes, and this solely depends on the NAS I/O capabilities.

What happens if the NAS I/O is not fast enough to let the checkpoint finish creating a new file for use with the chain, does it continue during the next interval and have a split backup file (2TB chunk first day and 1TB chunk the second day to make the full backup) or does the backup copy job start to fail indefinitely?

Yes, you can increase the copy interval, if having less frequent restore points is acceptable.

I was under the impression that if you set the backup copy job to include the actual job it would copy new restore points as they become available. If this is the case and I set the interval to 3 days wouldn't I still get all three restore points from the original job (just in a single incremental file on the backup copy side)?

[foggy]

What happens if the NAS I/O is not fast enough to let the checkpoint finish creating a new file for use with the chain, does it continue during the next interval and have a split backup file (2TB chunk first day and 1TB chunk the second day to make the full backup) or does the backup copy job start to fail indefinitely?

Transformation will just continue, regardless of the next interval start. Once it is complete, the job will perform its next synchronization cycle.

I was under the impression that if you set the backup copy job to include the actual job it would copy new restore points as they become available. If this is the case and I set the interval to 3 days wouldn't I still get all three restore points from the original job (just in a single incremental file on the backup copy side)?

Only a single restore point is copied within each synchronization interval.

[bg.ranken]

Transformation will just continue, regardless of the next interval start. Once it is complete, the job will perform its next synchronization cycle.

So I just want to make sure I'm reading this correctly. Does this mean that if the transformation cycle will take 5 days and the job is set to 1 day intervals that the job will not recycle itself and continue to let the transformation cycle to run to completion? Or will it recycle at the scheduled time but pick up where it left off? Or will it just fail indefinitely since it can't finish in the single day window?

[foggy]

Does this mean that if the transformation cycle will take 5 days and the job is set to 1 day intervals that the job will not recycle itself and continue to let the transformation cycle to run to completion?

That is correct.