2-Node Backup Cluster

A

akjome

New Member
Apr 22, 2025
9
0
1
Hey Folks.

We are looking into options for a two node backup cluster. We understand that we will need a quorum device.

So far, I have tested ZFS, iSCSI, StarWind. With two node cluster limitations, it seems that the best option so far is ZFS. Even though it does not have shared storage capabilities which would set us up for some data loss.

What we are looking for is the best choice for a two node cluster, that also provides shared storage, snapshots, and HA. This cluster would primarily be used as a standby cluster for when one of our production clusters goes down.

We are not opposed to adding a third node if a two node cluster is just not realistic. Thanks in advance for the advice.
 
Hi @akjome ,
A 2-node cluster with a quorum device is, functionally, a 3-node cluster. Just make sure the availability of the qdevice is treated with the same attention as the other two nodes.

As for your storage question - there’s no universal “best.” Since you’ve already determined that async ZFS replication is acceptable to you for your business case, that might be the “best” for your needs but not for someone else. Your best bet is to make a list of your must-haves vs. nice-to-haves, and compare that against your budget.

Good luck.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Last edited:
bbgeek17 said:
Hi @akjome ,
A 2-node cluster with a quorum device is, functionally, a 3-node cluster. Just make sure the availability of the qdevice is treated with the same attention as the other two nodes.

As for your storage question - there’s no universal “best.” Since you’ve already determined that async ZFS replication is acceptable to you for your business case, that might be the “best” for your needs but not for someone else. Your best bet is to make a list of your must-haves vs. nice-to-haves, and compare that against your budget.

Good luck.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
Click to expand...
With an actual three node cluster, would Ceph be a viable option? Or, would taking a node down cause the cluster into a read only state?

I like the features and functionality Ceph provides, though I read that it is more suited for five node clusters. I actually did set it up in a four node test cluster, though I know the ProxMox community prefers odd numbered clusters.

Background: We are looking at ProxMox as an option to replace our VMWare environment. The cluster that is currently running as a two node cluster has VSAN.

Thank you for the response, it is much appreciated.
 
I am not a Ceph expert, and I’m sure more experienced forum members will chime in. That said, yes, Ceph is an option with a proper 3-node cluster. It may not be optimal but could be good enough for your needs.

An even-member cluster is more likely to experience a "split-brain" condition than an odd-node cluster. However, in a properly spec’d business environment with solid networking, the chances of split-brain are lower. Still, it's always something to consider.

ESXi/VSAN is just different. You’re moving to a completely new solution, so don’t use your old environment as a direct comparison. If your business depends on the infrastructure, follow best practices for the platform you’re migrating to.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
You must log in or register to reply here.
Top Bottom