Backup repository suggestions? (iSCSI - NFS)

[xefil]

Hello to all,

I've some questions concerning choosing the best method to archive an efficient backup infrastructure.
Our setup is as follow:

• Veeam B&R + Veeam Enterprise manager on the same VM
• Veeam SQL DB on a different VM
• Production storage Netapp, backup storage Netapp as well (NFS for datastore - NFS and iSCSI for backup).
• Different Veeam Repository Windows and Linux

I would like to optimize the infrastructure. Actually I'm migrating backups away from local storage connected to the B&R to let the B&R do only as director for the backups, offloading it from backup data flow.
The doubt comes on the Veeam repos.
I'm introducing the scale out repo to extend the backups more in a dynamic way. The first introduction is the Backup from Storage Snapshot using NetApp. Here I've setup a physical server (Windows) with 3 connections: management, NFS to Netapp to our prodution VMs, iSCSI to our backup LUNs. Until here all works well and BfSS is great to reduce vSphere snapshot. The "problem" comes on the backup mount, in iSCSI. Netapp is known works not optimal on iSCSI and it's more proof on NFS. Even the storage usage, on NFS you have control on the content and the usage and stats are more readable. The iSCSI LUNS are a "black container" on Netapp. Even space reclamation is done via client and it's not visible on Netapp directly. The reclaim works not so good... it takes MANYMANY times and then it hangs....

So, if I move the physical Windows server to use NFS exports from Netapp instead of iSCSI, what disadvantages should I have? AFAYK the Instant Recovery needs the proxy act as NFS Server, so, the service (ports) would be exclusively used by the Veeam service, preventing to mount NFS shares on the backup Netapp. In case I DON'T use Instant recovery, it's enough to stop this service and mount an NFS share from my storage backup on this physical Proxy? Later I could then reinstall the proxy from Windows to Linux. I think it should work better.

For the other backups that are not using BfSS (due topology reasons) I could continue to use Linux Proxies with NFS shares, introducing Scale Out Repos.

What do you suggest about the scenario? Should I consider some other changes, do I missing something or seems reasonable what descrived?

Thanks a lot!!

Simon

[PTide]

Hi,

Production storage Netapp, backup storage Netapp as well (NFS for datastore - NFS and iSCSI for backup).

Are those two different NetApps, or the same one? If the former, then do you use backup NetApp as anything else than backup storage?

Thanks

[xefil]

Hello,

They are two different Netapp installations, two controllers each.
Production on FAS8040, backup on FAS2554.
We don't use Netapp SnapMirror between the two Netapp because not all VMs are under backup and for some we need Application aware backups (if it's that what are you asking, sorry).
The FAS2554 is used only for backup pruposes. Mostly on Veeam, but for some other legacy backups as well.

Thanks, Simon

[jmmarton]

The main consideration here is whether or not you want to use ReFS, assuming the Windows server used for the repository is Windows Server 2016. If you use iSCSI the disk is managed completely by Windows and can be formatted with any filesystem you choose. That's not the case with NFS so you would lose potential ReFS benefits.

Joe