Replication of VM and data consistency (ZFS)

[mephisto]

Hi guys,

I'm wondering how the data from a running VM replicated with consistence to another storage as I would expect some sort of guest interaction to trigger VSS for example to ensure there is no missing data due to writes being cached at the time the replication runs. I'm trying to compare this to Veeam on ESXi, where a regular Windows VM has VSS triggered, so no data is lost when the replication is done, the hypervisor and the VM are aware of the replication and we get a consistent replicated VM somewhere else.

Is this achievable with proxmox? What are the limitations?

 

[wolfgang]

Hi

Is this achievable with proxmox?

You must install the qemu-guest-agent in the VM. This is the same as the VMWare interaction tools(A part of it).
This daemon will handle the vss in the VM if a snapshot is made.

What are the limitations?

What do you mean with limitations?

 

[mephisto]

So ZFS replication in proxmox triggers a quierce of the filesystem inside the VM?

Does the backup job in proxmox also trigger the log flush for MS SQL and MS Exchange?

I would say limitations are like the item above, can the logs be flushed during backups? I'm directly comparing it to ESXi as it is what I use in production so I'm strongly considering switching to proxmox.

Thanks!

 

[sb-jw]

Does the backup job in proxmox also trigger the log flush for MS SQL and MS Exchange?

No, PVE and VMware don't do this, this is a Job for Veeam - thats not related to the HV itself. So you are able to use Veeam in your VMs too, but there is currently no HV Integration like Veeam and VMware has.

If you want to compare both System, then don't mix functions from third party tools with it. You have to compare VMware ESXi with PVE - without the Veeam functionality!

 

[mephisto]

At least if I can get crash consistent VMs replicas to another storage box that should already do the trick, then I can run a script from time to time to flush the logs from SQL and Exchange