Storage Reconfiguration Advice

[sixarm]

Hi there,

I'm just about to reconfigure my storage to hopefully improve throughput and also reduce any storage maintenance, such as rebuilds.

Current configuration is Windows Server 2016 physical server with 24 x 4TB 7.2K SAS HDDs in a single RAID6 volume as the Veeam repository, OS sits elsewhere. This is then formatted as one single ReFS volume. From there I have two Veeam Repositories pointing to the same volume, one for Archive (Copy Jobs) one for Backup.

My thinking behind the reconfiguration is that one single volume gives a single point of failure, lose the volume = lose everything. We recently extended this volume with additional HDDs and it took over a week to rebuild, which is bonkers. Also, if I split this into multiple RAID volumes (2 x 12 HDD) we should achieve better results allowing multiple jobs to run concurrently to each volume without a performance impact.

One thing I don't want to lose is the simplicity of having a single repository for backup and another for Copy Jobs. I'm thinking of creating a "Backup" and "Archive" repository on each volume, and then bang them together into a Scale-Out Repositories, letting Veeam take care of the placement. Would that be a wise move?

I'd be keen to hear your thoughts.

Thanks in advance!

[foggy]

Hi David, yes, you can do that. Just a little note, you will still have a single point of failure having both regular backups and copies on a single system (though different volumes), so my recommendation would be to consider some external copy of your backups as well (unless you already have it). Thanks!

[Alex Uryumtsev]

Hi David.

I'm thinking of creating a "Backup" and "Archive" repository on each volume, and then bang them together into a Scale-Out Repositories, letting Veeam take care of the placement.

I want to clarify. Are you going to create two Scale-out repositories for each type of the job(backup job and backup copy job)?

BR,
Alex

[sixarm]

Hi David, yes, you can do that. Just a little note, you will still have a single point of failure having both regular backups and copies on a single system (though different volumes), so my recommendation would be to consider some external copy of your backups as well (unless you already have it). Thanks!

Hi foggy,

Thanks for the reply.

I have another repository at our DR site that I use via Copy Jobs from Production. You're absolutely right, the Production repository server is a single point of failure, but at least we have copies of the data elsewhere.

I want to clarify. Are you going to create two Scale-out repositories for each type of the job(backup job and backup copy job)?

Hi Alex,

Thanks also for the reply.

I guess I will have to as you can't use a Copy Job with the same source and target repository can you? I came across this before, hence how I ended up with two repositories on the same volume in my current configuration.

[foggy]

Then your setup looks good, just be sure to assign this server enough resources and control the number of tasks it is about to process concurrently.

[Alex Uryumtsev]

I guess I will have to as you can't use a Copy Job with the same source and target repository can you?

David, feel free to use it this way. Technically, source and target can be the same storage.

[sixarm]

So here's what I've ended up with:

All repositories are set to 4 maximum concurrent tasks.

Logical Disk 1 - RAID5 (12 x 4TB) = D:\ - ReFS 64KB Block
D:\Archive = Archive Repository 1
D:\Backups = Backup Repository 1
D:\Replica = Replica Repository

Logical Disk 2 - RAID5 (12 x 4TB) = E:\ - ReFS 64KB Block
E:\Archive = Archive Repository 2
E:\Backups = Backup Repository 2

Scale-out Repositories
Scale-out Archive Repository = Archive Repository 1 & Archive Repository 2
Scale-out Backup Repository = Backup Repository 1 & Backup Repository 2

This Windows server also acts as a proxy. Max concurrent tasks is set to 12. Specification as follows:
CPU: 2 x Intel Xeon 6128 Gold CPU (6-cores @ 3.4Ghz)
RAM: 32GB

David, feel free to use it this way. Technically, source and target can be the same storage.

I'm sure I tried a Copy Job before with the target repository set to the same repository as the source and it threw an error that it couldn't copy to the same source/target?

[foggy]

I'd limit the number of proxy tasks to 10 due to the CPU cores requirement (1 core for each concurrent task).

[Alex Uryumtsev]

I'm sure I tried a Copy Job before with the target repository set to the same repository as the source and it threw an error that it couldn't copy to the same source/target?

That is wired. This configuration doesn't conflict with the system requirements. Please contact our support team, so they will have a closer look at your case.

Let us know how it will eventually work out for you.

[foggy]

I'm sure I tried a Copy Job before with the target repository set to the same repository as the source and it threw an error that it couldn't copy to the same source/target?

Having source and target as the same repository is not allowed, right. What Alex meant was the ability to use different repos on the same storage - this is fully supported.