single Virtual machine more than 16TB vbk

[cescoorru]

Hi to everyone. A customer of mine works with video editing and its fileserver is about 200TB already splitted in 2 Virtual machines, each virtual machine has 2 vmdk of about 51TB each [(51+51)+(51+51)] .
I made a backup job for each virtual machine and as soon as one of that reached the 16TB vbk's size the NAS stopped to write data. I really need that veeam implements a vbk split support because avery NAS I took a look won't support files bigger than 16TB.
I love veeam, and I hope some good reactions from team developer about this argument.
Have a nice day,
Francesco

[nmdange]

I hate to say this but NAS's really aren't the best choice for backup storage. You should get a physical Windows server with enough internal storage to hold your backup data, and format the backup volume as NTFS or REFS with 64k cluster size.

[cescoorru]

ok, so I think i'll connect that NAS via iscsi and i'll format it in NTFS by a windows machine... it looks like NTFS supports the max FILE size of 256TB? Is it real? I cannot find this information in the microsoft pages.... found it in wikipedia, I don't know if I can trust it.
So many thanks for your reply.
Best regards,
Francesco

[nmdange]

With NTFS, you should keep the volume size at 64TB max otherwise VSS and chkdsk don't work, and again make sure you format with 64K clusters. With the default 4K cluster size you'll still hit the 16TB limit. ReFS doesn't have such limitations, but you'll need to be on Windows Server 2016 with the latest updates from this month.

[csydas]

Echoing nmdange,

You can make NTFS sort of work for the huge file size you need, but it requires you format it with 64k Cluster Sizes and also format it with the /L parameter. You can see the cluster size information here: https://support.microsoft.com/en-us/hel ... -and-exfat

Format command syntax is here: https://www.computerhope.com/formathl.htm (Sorry, can't find an MS link and for some reason I don't think man format from my mac will help you )

But truthfully, it's a use case mismatch. You really do want a storage dedicated for this, not a file system repurposed to support this sort of file size. I'd never trust it, to be honest; you'd never had a chance to really know if the data is valid since you can only passively check it with things like chkdsk or third party tools for that. And you can just forget Veeam's health check, that'd never end. With something like btrfs, refs, and zfs, you get advanced storage tools meant to handle this sort of thing (integrity streams from refs, for example). If it were me, I'd be worried that I'd do it only to watch the backup go belly up because of a few bad sectors on one disk that I had no way of catching.

[cescoorru]

With NTFS, you should keep the volume size at 64TB max otherwise VSS and chkdsk don't work, and again make sure you format with 64K clusters. With the default 4K cluster size you'll still hit the 16TB limit. ReFS doesn't have such limitations, but you'll need to be on Windows Server 2016 with the latest updates from this month.

Ehi man, thank you very much for your tips. Now it is about 10 days that the backup is running without issues and it is incredible quick. I've passes my qnap LUN via iSCSI to a windows 2016 machine and formatted in ReFS 64kb. The lan connection is 10Gb and the average throughput is about 400MB/s . I'm very happy!
Best regards,
Francesco