Migrating from XCP-NG to Proxmox with Shared Storage and High Availability

wesleysilva01

New Member
Nov 10, 2025
4
0
1
Hey everyone!


I’m currently migrating from XCP-NG to Proxmox.


I’ll be running Proxmox with 16 servers in a cluster.


I have a 160 TB storage array (HUAWEI OceanStor V6, SSD NVMe) with two 10 Gb interfaces that will store all the VMs.


I want to use Proxmox with high availability so that when a VM migrates to another host, it only moves the CPU and RAM, since the disk will remain on the shared storage.


Should I use LVM over iSCSI?
Should I use NFS?
Which one is better?


I’ll also be using snapshots.
 
Hi @wesleysilva01 , welcome to the forum.

Both LVM/iSCSI and NFS are valid choices. NFS is generally more suitable for file operations and cross-host concurrent access. LVM/iSCSI is block storage, where concurrent access means different thing than the one in NFS.

There has always been snapshot capability with NFS. There is now snapshot availability in iSCSI case albeit as "technology preview", so may not be suitable for production use - depending on your risk tolerance.

You may end up using both options. The most common approach is place any shared files (ISO, QCOW, Snippets, Backups) on NFS export, while iSCSI is used for VM disks/images.


Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
I’ll be using it in a production environment. My main goal is to ensure high availability (HA) and stable operation, since I currently have over 100 VMs running. The ability to perform snapshots is essential for my setup.


However, live migration is also very important for me.
Do you know of any official documentation or guide I can follow to properly configure my environment?


Thanks!
 
The primary, migration oriented, document is https://pve.proxmox.com/wiki/Migrate_to_Proxmox_VE
All the best best practices references in multiple official documents are applicable. There is not a single official "XCPg to PVE" guide.

If you are running critical production, then you should probably avoid features marked "technology preview". In the end, only you can make that call.

Running QCOW2 over NFS with snapshots can induce some performance issues. I think there may be some ongoing development work in this area, but its not something I keep close tabs on.

If you are looking for official advice, you may want to engage a Proxmox Partner or Proxmox sales office.

Cheers

PS if you want assurances that your storage selection is perfect for Proxmox VE, you may also want to ask your storage vendor. Or, in case they are non-committal, you may want to re-evaluate your storage choices.


Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Last edited:
  • Like
Reactions: Johannes S
NFS is slower than iSCSI, but NFS is (really) easier to manage.

In both case you can increase the storage performance by:
  • using dedicated network interfaces for storage
  • using a separated VLAN for storage traffic
  • enable the Jumbo Frame.
 
  • Like
Reactions: Johannes S
This is an interesting topic for me, as we are currently evaluating Proxmox (in a dual HA cluster), and XCC-NG (in a dual HA cluster), both with shared storage. @wesleysilva01 I would be very interested in hearing your pros and cons of each system please.

 
This is an interesting topic for me, as we are currently evaluating Proxmox (in a dual HA cluster), and XCC-NG (in a dual HA cluster), both with shared storage. @wesleysilva01 I would be very interested in hearing your pros and cons of each system please.

I’ve been using XCP-NG for quite a long time. My environment has more than 200 VMs running across a total of 13 nodes, using NFS storage because at the time we didn’t have iSCSI available.


It has worked very well overall, and even with the limitations of NFS, we never had any issue that we couldn’t solve. One of the limitations of XCP-NG is that the community is smaller, so sometimes it’s harder to find solutions, although I was still able to resolve every issue I faced — nothing critical.


The main reason for moving from XCP-NG to Proxmox is the replication feature, which is something we need in our environment.


As for Proxmox, I can’t share any pros or cons yet, since I haven’t used it in production before. We are just beginning to evaluate it now.
 
This is an interesting topic for me, as we are currently evaluating Proxmox (in a dual HA cluster), and XCC-NG (in a dual HA cluster), both with shared storage. @wesleysilva01 I would be very interested in hearing your pros and cons of each system please.

 

It should be noted that several of the points in your table were disputed or corrected in the thread while you didn't bother to correct it. For this reason I consider referencing it as not very helpful FUD.