Single Backup job - Specify vcenter or VMs?

[jeffshead]

May have missed it but I could not find info for best practice when creating a new backup job. In particular, per vcenter, per host or specify each VM.

I guess the benefit of selecting the highest level container (vcenter) is that it makes the backup job dynamic so that any new VMs are also backed up without having to edit the backup job.

I have two ESXi hosts. Most VMs are on ESXi-1. ESXi-2 hosts a Veeam B&R server VM and the vcenter VM. I need to exclude the Veeam B&R server VM and the disk that contains the repository which is attached as the D drive to the SRV-FILE VM on ESXi-2.

Do you select the vcenter server, hosts or do you add each individual VM to a single backup job? Does it matter and what do most people do when they set up a job?

If I add the vcenter server, do all disks get backed up or will the exclusion I set (depicted in the image below) supersede?

Is this the correct way to exclude that disk? If I attempt to add a disk exclusion to the vcenter line, I'm presented with long lists of IDE, SCSI, SATA and NVMe disks. I don't know what to include/exclude since I do not have all of those disks that are listed. Do I tick them all and untick only the one that I want excluded? Then remove that second line for the SRV-FILE VM? Is that the proper way to do this?

Should I create a separate backup job for each host? Probably no more than 20 VMs combined.

[Mildur]

You can configure the backup job with vcenter as a source, if you want to have automatic included all vms you create. If you configure single vms in the job, you need to configure the job when you created a new vm.

I recommend to configure your backup jobs to use „per vm backup files“, if you on the enterprise or enterprise plus license. If one backup chain is broken, it would concerns only a single vm. This can be done in the backup repo Settings.

https://helpcenter.veeam.com/docs/backu ... ml?ver=110

You can go up to 300 vms per job if you want to follow best practice.
Please have a look at the best practice Guide:

https://bp.veeam.com/vbr/VBP/4_Operatio ... ckup_jobs/

Exclusion rules superseed the „include all“ line from the vcenter. You are good to go.

Disk exclusion is done by disk type and ID of the disk connection. You will find the correct number in your vm properties in the vcenter. Please leave vcenter on the first line with All disk configured. Or it would exclude something on every vm you want to backup.

https://helpcenter.veeam.com/docs/backu ... .html#disk

[foggy]

Just a side note, something to keep in mind regarding the setup in general - storing backups on VMFS is not considered the best practice for several reasons.

[jeffshead]

...storing backups on VMFS is not considered the best practice

OK... back to the drawing board. For my use case, what should I do?

I have two ESXi hosts. Both use internal storage for VMs. One host has an additional 16TB RAID10 array (12 HDD) that will be used for file storage and the Veeam repository. How do I attach this array to the file server VM and also use it for the repository if I do not make the array a datastore (VMFS)?

[Mildur]

You backups shouldn't be on the same storage array as your productive data.
If it‘s a dedicated storage array for backups, use iscsi to directly connect to it from a vm.

If itˋs FC Storage, then you can‘t use it with best practice in mind.

Not a very good idea, to store fileserver data and backup data on the same array.
What if a hacker or insider destroy the raid on your array? Your productive data and backups are gone at the same time

[jeffshead]

Not a very good idea, to store fileserver data and backup data on the same array.

I totally understand this and that my setup is crappy but that array is the only storage that I have that is large enough to accommodate both file server storage and Veeam backups.

This was a Windows physical server that I converted to a second ESXi server. I think I should have left it as physical. I also used to perform a separate backup of the entire server, with Veeam agent, to an attached USB 3.1 caddy with a large HDD. Took about 8 hours but that gave me a complete backup of my files and my Veeam backups.

Now that the server is virtualized, I tried to recreate the same scenario with USB passthrough and a second backup job using the Veeam B&R VM but the backup to the USB device is way slower than when it was a psychical server. It's been running for over two hours and its only at 6% so what used to take 8hrs will now take 33hrs?

Physical machine backup using Veeam agent:


VM backup using Veeam B&R VM:


Would PCI direct passthrough of the USB controller make any difference? Is it much faster than USB device passthrough? Also, would it be better to install the Windows agent on the file server VM and use it to back up to the USB drive instead of using the Veeam B&R with the USB device as a second repository?