Is Replication coming on non-ZFS storage ?

[DC-CA1]

Hi everyone our datacenter is built on shared sas storage between blades servers.

Really solid but we lack of replication that we do have with same type of storages on hyperv Custers.

Is there anything coming in a next release of proxmox or it will never appen..?

 

[RolandK]

i guess that qemu does not contain appropriate block level replication feature, so that's why zfs is being used for this and why it's not possible with other type of storages.

also see https://forum.proxmox.com/threads/replication-with-qcow2-files.51256/

did you even know that qemu has some feature like vmware fault tolerance, called "colo" ? https://wiki.qemu.org/Features/COLO
not sure about the current state, i.e. if it's production ready.

 

[bbgeek17]

@RolandK mentioned ZFS replication already. It's something to be aware of, but it wouldn't apply to your existing shared-SAS storage. The other supported method is via Ceph, which is also not applicable to your current storage. As @RolandK said, the storage agnostic approach would be the only way to plug into your existing environment. But, I'm not sure that's on the roadmap.

We've also been developing a DR/Replication solution that will work with Proxmox. It should be available early next year. Our main objectives are maintaining high performance, providing replicated storage supporting a Proxmox site transition, and optimizing the data transfers. Again, I don't know if this helps you, but ping me if you want to chat.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[Stoiko Ivanov]

if you can pass through the same LUN to multiple blades the classical approach would be putting LVM (not thin - so no snapshots) on a LUN and add that as shared storage to all your blades - that way all your nodes have access to the VM's disks.

 

[spirit]

Technically, I think it could be possible to do continous replication with qemu directly, without snapshot, like for backup, but always running. With using the new fleecing backup from qemu 7.0
https://wiki.qemu.org/ChangeLog/7.0#Block_devices
with a buffer disk to handle the fast write. This allow to create some kind of virtual snapshots.

But yes, it's a lot more complex to manage than sync snapshots from storage. (could be great to have ceph rbd replication too, mainly when remote clusters will be implemented)