Proxmox and IBM Flashsystem 7300

tedew

New Member
Nov 13, 2025
4
0
1
Hello,
I have a cluster of five hosts running vSphere. These five hosts are connected via a SAN network to an IBM FS7300 storage system(Fiber Channel).
All virtual machines are stored on the IBM storage.

My general question is:
Would it be possible to migrate this infrastructure to Proxmox VE 9.0 and keep more or less the same functionality we currently have with VMware? (Snapshots functionality is mandatory for us.)

BTW,
I’ve read some posts, but most of them are from before version 9.0.

BTW,
Sorry for the maybe trivial questions.
 
Hello,
yes, we moved our older but still valid vSphere 8 licenses there, but as far as we know, starting from vSphere 9 we have to purchase VCF or VFF — which has a big impact on costs :)

So, we’ve started an internal discussion to plan for the future.
 
 
I do not own an IBM FlashSystem but according to the documentation, you can migrate to PVE 9, while keeping your existing features.

Firstly, you have to configure FC on the PVE hosts directly from the Linux CLI.

Secondly, you have to choose a backend over your SAN
According to the PVE backend matrix you should try ZFS-over-FC or else LVM-over-FC
Do not worry about thin provisioning within PVE since your IBM FS7300 already handles it natively.

Snapshots
Unlike VMware vSphere VAAI, there is no PVE plugin yet to use native storage snapshots on IBM Flash System. However you can
  • use Proxmox snapshots. It is slower but it always work
  • use snapshots from your backup tool. For example Veeam supports both Proxmox VE and IBM FS native snapshot.
According to grok.com, only few storage vendors already support Native Storage Snapshots for PVE 9 yet.
As you can see, most of them use either LVM-over-SAN or ZFS-over-SAN
  • Blockbridge with LVM-over-NVMe
  • TrueNAS with ZFS-over-iSCSI
  • Netapp (ONTAP) with both NFS or ZFS-over-iSCSI
  • Pure Storage (FlashArray) with LVM-over-iSCSI or LVM-over-FC
  • HPE (MSA 2060) with ZFS-over-iSCSI or ZFS-over-FC)
Personally, I use Netapp with NFS and PVE snapshots for a 3 -nodes cluster because it is easier to manage.
 
Last edited:
According to the PVE backend matrix you should try ZFS-over-FC or else LVM-over-FC
Officially, there is no such thing as ZFS-over-FC. You may be thinking that it is similar to ZFS-over-iSCSI, however it is not.
The built-in PVE ZFS-over-iSCSI plugin not only internally provisions ZFS volumes inside the backend storage, but also exports them via iSCSI. I am not aware of any plugin out there that does FC exporting/mapping/zoning. And there are very few, if any, SAN devices that use ZFS and provide FC connectivity.

use snapshots from your backup tool. For example Veeam supports both Proxmox VE and IBM FS native snapshot.
It is unclear what you mean by "Veeam supports". The Veeam backup plugin for PVE does NOT use storage snapshots in any way. This is very different from many other Veeam supported infrastructures where Veeam directly invokes storage snapshot functionality.
As you can see, most of them use either LVM-over-SAN or ZFS-over-SAN
ZFS-over-iSCSI is a very specific scheme. Personally, I think it is misnamed, but it will stay like this for historical reasons. The approach creates a ZFS volume and exports it as raw device via iSCSI. So it is really iSCSI overlaid on ZFS.
According to grok.com, only few storage vendors already support Native Storage Snapshots for PVE 9 yet.
As you can see, most of them use either LVM-over-SAN or ZFS-over-SAN
  • Blockbridge with LVM-over-NVMe
  • TrueNAS with ZFS-over-iSCSI
  • Netapp (ONTAP) with both NFS or ZFS-over-iSCSI
  • Pure Storage (FlashArray) with LVM-over-iSCSI or LVM-over-FC
  • HPE (MSA 2060) with ZFS-over-iSCSI or ZFS-over-FC)
Luckily for humanity the AI is not always correct, yet. Netapp's WAFL is most certainly not compatible with ZFS-over-iSCSI. Neither is HPE SAN - there is not a shred of ZFS in that box.

Finally, there is no LVM or any other sort of volume manager or file system layer in our integration with PVE. We present Raw disks (iSCSI and/or NVMe) and offload snapshot functionality back to the storage where its most efficient.

Cheers


Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
  • Like
Reactions: Johannes S
  • TrueNAS with ZFS-over-iSCSI
  • Netapp (ONTAP) with both NFS or ZFS-over-iSCSI
  • Pure Storage (FlashArray) with LVM-over-iSCSI or LVM-over-FC
  • HPE (MSA 2060) with ZFS-over-iSCSI or ZFS-over-FC)
dont trust AI slop ;) only Truenas has a plugin available for zfs over iscsi afaik, and its not provided or supported by either Proxmox or Truenas. The rest dont expose the underlying filesystem (and is not zfs at all.)
 
  • Like
Reactions: Johannes S
You can migrate your infrastructure to Proxmox VE 9, but since you mentioned that snapshot functionality is mandatory, there are some important considerations.

You basically have two options:

1/ Use Proxmox 9 with snapshots on iSCSI/VM using QCOW2 disks.

This is possible, but QCOW2 snapshots are still considered experimental and not fully production-ready.

There are performance issues, especially with large disks, and risks of data corruption with chained snapshots.

For very large disks, there is a possibility of metadata corruption in case of power loss or unexpected VM/KVM crashes.


2/ Use RAW disks (iSCSI LUNs / LVM) and manage snapshots on your SAN.

Proxmox itself won’t manage snapshots; your SAN (like IBM FS7300) handles them safely. This is the most reliable method, especially for large disks.

From my research and experience:

There is no supported iSCSI-over-LVM or ZFS-over-LVM solution from vendors like HPE or others.

The only supported approach for ZFS snapshots over iSCSI is via a TrueNAS plugin. As far as I know, this is not officially supported in Proxmox 9.

If you want Proxmox-native snapshots with reliability for large disks, solutions like Blockbridge might be worth exploring.

So, yes, migration is possible, but if snapshots are critical, I strongly recommend keeping snapshots on your SAN or considering a specialized solution, rather than relying on QCOW2 snapshots in production.
 
  • Like
Reactions: Johannes S
Hello,
Thanks for all repiles.
Our vSphere contract will expire at the beginning of 2027. Do you think it’s better to wait and see if Proxmox will receive more features or improvements next year ?, especially regarding snapshots and better LVM-over-FC support?
 
I would expect the snapshot feature of PVE9 to be more mature then. Anpther question is whether you want to use dedicated storage Hardware after 2027. How long are you planing to use your IBM Flashsystem?
If you would need to renew your storage Hardware in 2027 anyhow it might je worth to seitch to hardwarw working with ProxmoxVE ( e.G. a NFS filer like Netapp or Blockbridge ) or go the hyper-converged-infrastructure ( Ceph).
In general it's recommended to prepare a migration early, so maybe you can setup a Lab to test the options ( Snapshots as in PVE9 with your IBM, Ceph, different storage hardware). Two years are a long time to get fammilar with the available options in PVE and you don't need a support subscription for it ;) It might also be worth a shot to get some consulting from a Proxmox Partner which approach might work best for you