What are people doing for cross-site replication?

[xpaz]

I know of only a few options;

- Not cross site, but Proxmox natively ZFS replication
- Storage level replication, eg ZFS, or natively through the appliance like Pure Storage or perhaps something like Starwind VSAN
- Ceph to Ceph - https://github.com/lephisto/crossover

Not considering Proxmox Backup Server, Veeam Backup & Recovery or Vinchin Backup as they are exactly that, Backup, not DR.

I have tried rsync before, however it struggles to only transfer incrementally, it often will transfer the entire VM disks.

Wondering what others are doing in this scenario..

 

[Alex (StarWind)]

Hey,

It depends on your network between sites. StarWind VSAN can do synchronous replication between your sites. Synchronous replication requires network latency less than 5 ms. https://www.starwindsoftware.com/system-requirements#vsan

If your network has higher latency, ZFS replication is a good choices. Configure replication jobs to replicate data between the nodes on schedule:
https://pve.proxmox.com/wiki/Storage_Replication

Best regards,
Alex

 

[wchesley]

Veeam Backup & Replication can handle disaster recovery, it's one of their core features. You'll need VBR at both locations and setup replication jobs between them. This is how I would do it for enterprise for support and reliability reasons. At work, we do this for a few clients using Vmware infrastructure at both locations. We use Veeam Service Provider Console to manage the backup servers in our datacenter and client locations. I haven't had a chance to set this up with Proxmox on both locations, as I only run Proxmox in my homelab. If I were to set up a DR site for my homelab, I would use ZFS replication jobs between the sites.

 

[VictorSTS]

Not considering Proxmox Backup Server, Veeam Backup & Recovery or Vinchin Backup as they are exactly that, Backup, not DR.

Wait until your disaster is a ramsomware, an intrusion to your main site or some crazy corruption at the storage level an you'll start to wonder why you didn't use some kind of remote backup solution that you can recover from. Storage replication will cover some disasters, but there are others that require some other approach. In most cases having a remote PBS syncing from the local one and doing a restore when disaster happens is more than enough, but obviously that depends on RTO/RPO and budget.

 

[xpaz]

Veeam Backup & Replication can handle disaster recovery, it's one of their core features. You'll need VBR at both locations and setup replication jobs between them. This is how I would do it for enterprise for support and reliability reasons. At work, we do this for a few clients using Vmware infrastructure at both locations. We use Veeam Service Provider Console to manage the backup servers in our datacenter and client locations. I haven't had a chance to set this up with Proxmox on both locations, as I only run Proxmox in my homelab. If I were to set up a DR site for my homelab, I would use ZFS replication jobs between the sites.

Veeam does not support Replication for anything other than VMware or Hyper-V. Backup support seems unstable when I tried it on the latest build.

Wait until your disaster is a ramsomware, an intrusion to your main site or some crazy corruption at the storage level an you'll start to wonder why you didn't use some kind of remote backup solution that you can recover from. Storage replication will cover some disasters, but there are others that require some other approach. In most cases having a remote PBS syncing from the local one and doing a restore when disaster happens is more than enough, but obviously that depends on RTO/RPO and budget.

We already have a backup in place, that's why I'm asking about replication for DR scenario.

Hey,

It depends on your network between sites. StarWind VSAN can do synchronous replication between your sites. Synchronous replication requires network latency less than 5 ms. https://www.starwindsoftware.com/system-requirements#vsan

If your network has higher latency, ZFS replication is a good choices. Configure replication jobs to replicate data between the nodes on schedule:
https://pve.proxmox.com/wiki/Storage_Replication

Best regards,
Alex

This seems to be where we are headed. Storage level replication.