Backup Copy Jobs with large VMs

[Amyd]

After updating to Veeam 7, I am trying to change our secondary, off-premises backup from direct backups to backup copy jobs, since we have to minimze the impact of the backup on performance within work hours. We have about 36 TB of data to backup, most of it concentrated within two VMs with about 12 and 20 TB, and the off-premises backup target is a portable iSCSI Synology unit, which will be connected once a week during these work hours for the incremental backup.

I am, however, having a bit of a trouble getting the first full backup to work - obviously, it will take quite a few days for the full backup to sync from the primary backup repositories to the secondary one and the Backup Copy job keeps breaking up. It first broke off when the primary backup jobs ran, and upon 3 retries it seemed to have given up on resuming the transfer from where it ended. I deactivated the primary backup Jobs - ok, fair enough, I thought, maybe it didn't like having the full backup changed (the Primary Jobs are reverse incrementals). However yesterday the backup Job broke off again, once the interval for the backup copy job was reached (set to the maximum of 1 Day at 00:00), and I am not quite sure what will happen once it fails 3 times (besides having a bit of a problem with leaving the primary Jobs deactivated so long).

Am I overlooking something simple about the Backup Copy jobs, or thinking about it the wrong way?

[veremin]

Have you tried to use existing backup file as a seed for backup copy job? Should do the trick.

Also, I’m wondering what type of backup copy job you’re going to use. Direct or WAN Accelerated one? Please be aware, that with the current implementation it’s not possible to use WAN Accelerator feature for VMs that have disks that are larger than 1.8 TB. Though, this issue should be fixed soon.

Thanks.

[Amyd]

Thanks, I actually didn't think about that idea, but I am now also not sure if it would work in our case - we actually have two primary backup repositories, which each backup about half of our VMs, so we have two backup chains. However, on the secondary backup I'd like to combine them in a single backup chain, I am guessing it won't work if I just copy the two full backups files, Veeam won't know to combine them in a single one, right?

I guess I could do two backup copy Jobs to mirror the primary ones, bit more management, but OK. I assume the procedure for seeding would be to copy the full backups from the primary repository to the secondary, so this would mean I would still have to leave the primary backup jobs deactivated for the period, right? Also, I wouldn't benefit from better deduplication between the jobs, correct?

The Backup Copy Job is a direct one, since the "off-premises" backup target will be in fact on-premise for the time it takes to do an incremental backup during work-hours, directly connected to the iSCSI LAN.

[veremin]

However, on the secondary backup I'd like to combine them in a single backup chain, I am guessing it won't work if I just copy the two full backups files, Veeam won't know to combine them in a single one, right?

Correct.

Also, I wouldn't benefit from better deduplication between the jobs, correct?

Backup copy job doesn’t copy backup files as the whole, instead, it synthetically creates required restore points in remote location from VM data in source backup repositories. Therefore, we shouldn’t talk about deduplication between backup jobs, but rather about deduplication between Virtual Machines.
Though, you were right that that there would not be any deduplication between VMs located in backup copy job#1 and VMs located in backup copy job #2.

I recommend you creating a new backup job and backing up within it Virtual Machines that are nowadays backed up by existing backup jobs. Then, you can use resulting backup file as a seed for backup copy job. Not only will this scenario allow you to have all VMs within one backup copy job, but also it will guarantee better deduplication rate.

For the information regarding backup copy job seeding, kindly, refer to this thread.

Thanks.