Deploying new backup storage, looking for file system recommendations

[kapple]

Hi Veeam Community,

We have purchased a NetApp AFF-A250 array to replace our old, out of support Drobo arrays to use as backup repositories.
Our Veeam B&R server runs Windows 2022. It is used to backup VMware guests, and a handful of Windows agents.
I am looking for recommendations regarding what file system to use once we have the NetApp array racked and connected to our network.

Currently our Drobo backup targets are connected via iSCSI to our Veeam servers, and are formatted as NFTS volumes.
My thought was to share out the yet to be created NetApp storage as NFS, and use that as a backup repository, or carve it into multiple NFS shares to create multiple backup repos, so we can have multiple jobs run at once. I am not sure if this is the best way to go for performance or not.

There will be one other difference in how we backup our VMware VMs. We had been backup up VMs with jobs we created with similar VMs grouped together (Windows DB servers, Windows file servers, Linux servers....), and using per VM backup chains. Since we have an opportunity to recreate all our jobs, and now that we also have new VM storage (another NetApp array), I am toying with the idea of backing up from storage snapshots. This is something we have never done before.

So, my question is, are there either any Veeam best practices, or recommendations you all may have regarding the best connectivity and filesystem to use for our new backup repositories.

Thanks in advance for taking the time to consider this question.
- Ken

[tyler.jurgens]

In your environment, I'd go with an iSCSI LUN formatted with ReFS. That would give you better performance and block cloning ability. You could attach them directly to your Veeam servers as you do today with your Drobo.

However, if you have the option to modify your setup a bit, I would suggest you deploy a Linux server. Mount that NetApp again as an iSCSI LUN and format it with XFS. That way you can setup a Linux hardened repository. You get immutability, plus the same block cloning ability as ReFS. I have also personally found far better stability with an XFS repository.

I would avoid NFS, and preferably also avoid an SMB share.

[pybfr]

I don’t really think it is serious to plan a déployment in this day and age without immutability. So, like Tyler said iSCSI with a Linux XFS server, or an S3 bucket.
I’d stay away from anything else (NFS, SMB, ReFS)

[kapple]

Thank you Tyler and pybfr for your feedback on this issue, I appreciate it.