Promox Replication DC to DRC
- Thread starter rodhu
- Start date
First off, don't create 1 cluster of all 6 servers, but 1 cluster per datacenter with 3 servers. With all 6 in one cluster you'll lose quorum if 3 fail, and can't use your DR, so you'll want to avoid that of course.
What kind of storage are you looking at / planning to be using, because that will be one of the main contributors / limiting factors of what you can use.
What we for example use (maybe not optimal, but we had to re-use hardware we already had) was a iscsi storage-server that syncs the storage for us, and a script that backups over the vm-config to the DR-location's storage so that we can load that into our other servers and start running there if we need it.
What kind of storage are you looking at / planning to be using, because that will be one of the main contributors / limiting factors of what you can use.
What we for example use (maybe not optimal, but we had to re-use hardware we already had) was a iscsi storage-server that syncs the storage for us, and a script that backups over the vm-config to the DR-location's storage so that we can load that into our other servers and start running there if we need it.
Last edited:
if we are using san storage and 1 cluster per datacenter with 3 servers what type of replication suggested? example zfs, replication
Check the featureset for the SAN if you have more than one from the same vendor (may also need to be from the same series/model) they do sometimes offer replication features within the san itself.
At which point you'd then need to replicate the configuration between the proxmox clusters.
Depends on your requirements on this, yet we use PBS to get the data from a local PBS to the remote PBS and we restore the VMs regularly. There is currently no other way to this incrementally without ZFS, which does support incremental send/receive.
ZFS cannot be used locally in the cluster easily, because you will have a complicated setup that does not scale well.
Yet if you would have ZFS on your SAN, it would work. This is the same approach as @dragon2611 talked about.
Yes, ZFS and CEPH are both capable of doing incremental replication, yet both have a totally different local setups: ZFS is only single node and CEPH a clustered setup with at least 3 nodes.