Backup one by one

[GianlucaCroci]

it's possible to create a Backup Job that do a Backup one by one? In one or more cases I've Servers that need to be backuped in sequence.
I know that's possible to limit the maximum concurrent tasks on the Storage and on the Proxy too. Another way?
I don't know if with our license "Enterprise" it's possible.
At the moment I created separate Jobs with only one Server per Job, which I follow in the chain. Perhaps there is a better solution (example with scripts), more simple to manage.

Thanks a lot for the help.
Kind regards

[Shestakov]

Hello Gianluca,
Chained jobs and limited number of concurrent tasks are working methods. What is in them that you do not like?
There is no such a special option and license edition doesn't make any difference here.
By the way, do the VMs to backup have equal amount of VM disks?
Thanks!

[ferrus]

I had to do this for VMs in an Exchange DAG. Despite a number of different methods, there really wasn't a perfect way.
The best way I used to achieve the result was chained jobs - something that's rarely recommended here.

The problem for me, was I was trying to avoid concurrent VM snapshots across the DAG, which even with SAN-storage integration and limiting concurrent tasks on repositories etc - could still occur at the same time.
It would just prevent the actual data from being backed up concurrently - which actually, wasn't a concern for us. Just the timing of the VM snapshots.

I don't know which situation you're trying to avoid, but chained jobs could still be the best solution.

[Shestakov]

Agree, because proxy concurrent tasks work per VM-disk, so you either set = 1 and job will take long or >1 and several VMs will be backed up simultaneously.

[skrause]

Chained jobs are a perfectly valid solution for use cases where you don't want snapshots to overlap. They aren't recommended for general scheduling use as the Veeam job scheduler is more efficient at managing resources.

Keep in mind that the chained jobs work on job "finish" not "success" so if the first job in your chain fails and it is set to "retry 'x' times on failure," you can have a situation where the second job in your chain is running at the same time as the first job is retrying.

[ferrus]

Correct.

Also - it covers the entire job, not just the backup portion.
So if you have a 30 minute incremental backup, followed by a 6 hour merge/compact/corruption check, the second chained job doesn't get started for the whole 6h30m, despite the window being available for 6 hours.

There' really isn't any perfect option, unfortunately.

[skrause]

In the case of backing up things like DAGs, DFS, or SQL Always on replicas there is always the option of just not backing up the entire VM on the "copies". Provided all of your mailboxes/files/databases are on every server, just backing up one for DR or compliance purposes would fulfill the 3-2-1 rule because it is really the data that is important not the "machine" it lives on. You can do a "system drive only" backup of the machine to cover for OS level issues which would not snapshot your data volumes if you need that.

Your internal policies may differ, of course. But I just wanted to throw that out there as food for thought.

[ferrus]

Actually, the eventual solution for us - was to use Veeam Agents instead.

We still use VM backups for the SQL Always On groups, since they're much less sensitive to failovers during the VM snapshot stun, but couldn't make the Exchange backups work consistently and truncate the logs.

With the Agent backups - we haven't looked back. Everything has worked perfectly since day one, faster than the VM backup, and not a single failure.

Perhaps Gianluca can advise us what type of server, he needs to back up separately?

[GianlucaCroci]

Godd Morning,
thanks a lot for all replyes.
My "problem" is that we've few VM (DNS primary and secondary, Exchange, SQL Cluster, ...) that need to be backuped one by one. I can do a chain jobs, but I think is more easy to manage with a single job (one for DNS, one for Exch, one for SQL).
If there aren't Others solutions, I think that I'll create a new Proxy Server with only one "Concurrent Task", and associate the jobs thats need this type of backup at this Proxy.
What do you think? Possible problems you see about this?

thanks again, and have a god day
kind regards

[Shestakov]

The main issue is that's going to take long, especially if you have several disks on each VM.
From the other hand you can give it a try and see if it works fast enough.
Thanks!

[GianlucaCroci]

Good Morning,
I understand. So it's better I check how many Servers I've to Backup one by one, and see how many Disks they've. If the 70-80% have only one Disk and I can save 3-4 Jobs, I'll install a new Proxy with only one "Concurrent Task", otherwise I'll continue with Jobs chain.
Thanks so much for all answers.
Kind regards

[Shestakov]

You are welcome, Gianluca.
Hope it will work well for you.