Feature Request: Option to prevent certain servers backing up at the same time

[pber]

I would like a feature where I have many servers in the same backup job, but flag specific servers to not backup at the same time. So only those specific servers would be prevented from performing the backups at the same time. Similar how you prevent the Veeam server itself from backing up until last.

We have some servers of high importance that we don't want actions done to them at the same time to reduce risk. e.g. Domain controllers, Web Front ends, DB servers, etc.

We currently get around this by scheduling them in separate jobs, but they still have the potential to run at the same time if we have a long running job that spills into the next backup window.

[foggy]

Hi Patrick, what kind of risk are you trying to reduce?

[pber]

We have a critical industrial control environment that we have servers that run in a cluster. On occasion we’ve had unplanned failovers that have occurred for various reasons: high CPU, network usage, disk usage, high memory, etc. We have also had failovers caused by the VSS process during a Veeam backup. For example, If ServerA was caused to fail over to ServerB or ServerC due to a VSS issue during a backup, we don’t want ServerB and ServerC to be at risk of having a VSS issue at the same time it is trying to become the active server. We also have active/active clusters that we don’t want the extra cpu, disk, or network burden during backups to be occurring at the same time. We have this concern for both VMs as well as physical agent backups

As I mentioned in my previous post, we get around this issue with scheduling our critical servers separately. However, with around 500 servers it is tough to get them all backed up within our backup window every day. On occasion a job might run longer than normal and bump into the next scheduled backup. Normally this isn’t an issue, but if it bumps into our critical jobs is puts us at risk of having potential failover issues. I’ve tried to use the High priority option, but it still doesn’t get around that issue as I would have multiple high priority jobs that may run at the same time.

If I had an option to choose specific servers in a job that never run at the same time (Similar to how the Veeam server itself waits until all other backups in the job are done before it backs itself up), it would reduce our risk if a failover does occur during a backup that the other servers aren’t also being backed up. Even if one of my other jobs run into this job, it would still never have simultaneous backups of my critical servers.

[PetrM]

Hi Patrick,

What about to chain the separate jobs with critical servers using "After this job" option to get rid of simultaneous backup? Also, I would recommend to investigate the issue with failover during VSS processing as it does not seem to be expected.

Thanks!

[soncscy]

Just my 2 cents,

I think I know about the kind of industrial control systems Patrick is talking about, and it's not uncommon for them to be extremely sensitive to IO pressure of any sort (snapshot/diskshadow both can be too much). It's not really something you can tweak typically as the systems are meant to be for like factories and such controlling tons of machines and they're incredibly sensitive to any interruptions.

I'd approach this with scripting and tags. It's a bit complex, but I'm not sure there's a way to effectively do what you're meaning to do as I don't think Veeam's scheduling has a concept of "don't run" scheduling" besides the backup window.

I guess you could have some "monitoring" script that just watches and groups VMs (as you see fit somehow) and checks for which ones are being backed up and either disables or enables the job, or sets an exclusion for the active VMs.

I'm not sure how many critical machines you're dealing with and your preferred default behavior, but I think this is about the only way to handle it.