Which is better - multiple servers in single job or multiple jobs?

[NorthGuard]

I am reconfiguring one of my Veeam setups that is backing up about 8 servers.

I am curious which is the preferred method for configuring the backup jobs:

Create one or two jobs that contain multiple servers
OR
Create individual backup jobs for each server.

[chris.childerhose]

This is a broad question, as you have not posted how your Veeam environment is set up. Typically if you have 8 servers and depending on your setup, you can use just one job for those. Other things like your SLA, and RPO/RTO will also affect how you set up your jobs.

[NorthGuard]

When you mention we can just use one job for those, do you mean a single job containing all servers?

Also, what was SLA and RPO have to do with how many jobs are created? Recovery procedures would essentially be the same if they were each their own job vs a single large job, correct? Or am I missing a piece of the puzzle?

FYI on the environment:
VM B&R Server in separate network
QNAP NAS in separate network
note: separate network is in same building and separated from LAN via hardware switch + firewall rules

[chris.childerhose]

Yes, you can have one job with all servers contained in that job.

SLA and RPO depending on how you recover, can mean separating, say, a SQL server versus other general or file servers. I work for an MSP, so we have SLAs and RPO/RTO for our clients so, which dictates how we set up jobs and how many servers are contained in the job. If this is just local, then you are correct about recovery, and these would not factor into your setup.

[NorthGuard]

I am still not following how SLA and RPO play into the job setup. The recovery procedures would be the same if it is one job vs many jobs....

[chris.childerhose]

The SLA is for your customers, so if they have one with you, that will dictate how jobs get set up, which includes the requirement for RPO (Retention) and RTO (Restore Time). I'm not sure how much clearer I can be as we have these with our customers (I work for an MSP), and we sometimes set up jobs differently. If that does not pertain to your environment, ignore my statement.

[Net Runner]

Theoretically speaking, with separate backup jobs for each server, you might have better granular control over the backup process and more flexibility. This is nice for 8 machines, and you can imagine that this will be hard to manage as soon as you have more VMs. The best way here would be just simple testing. Create a single backup job, do the full backup, and measure the time, performance, test restore, and how you feel about it, and do the same with separate jobs. Compare and go with any config which is better for you. Remember that in case, your proxy server might be a bottleneck.

[newman]

If you have multiple jobs and the backup repository is not a dedup appliance, take into account dedup/compression is within the job. So if you have two jobs, no dedup between them.

[MartinO]

I am still not following how SLA and RPO play into the job setup. The recovery procedures would be the same if it is one job vs many jobs....

You may have one server which requires a 4 hour RPO, while your others are only 24 hour. So unless you also want to backup your other 7 servers every 4 hours, you'd need at least two jobs.

[JeroenL]

I am still not following how SLA and RPO play into the job setup. The recovery procedures would be the same if it is one job vs many jobs....

RTO isn't really depending on job settings. It mainly depending on the performance of the backup environment and network as a total.
With Instant Recovery it is possible to create an RTO as low as a few minutes. How many VM's you can run with Instant Recovery is mainly limited by hardware.

RPO is VERY depending on job settings. A RPO of one hour can only be realized if you run you backup task at lease one time every hour.
So the job schedule is the primary parameter that defines the RPO of the data within that backup job.

If you have 8 Windows servers with a RPO of 24 hours and all backups can be created within the same backup window. I would create one job starting after production hours using Application Aware Processing.
If you have a mix of Windows and Linux servers with an RPO of 24 hours and all backups can be created within the same backup windows. I would create two jobs. One job containing all the Windows servers with Application Aware Processing enabled and a second job linked to the first job containing the Linux servers with Application Aware Processing disabled, VMware Tools quiescence enabled and maybe a pre-/post script to stop and start applications running.

The more requirements for specific servers, the more separation I use in my job creation.
A SQL Always on cluster with Transaction log shipping gets its own job. This gives me maximum flexibility with scheduling and fine tuning options.

I hope this post helps.

[chris.childerhose]

Thanks, everyone, for chiming in on the RTO/RPO piece and explaining further. Great to see the community do this.