On Thu, Dec 13, 2018 at 5:01 AM Zhenshi Zhou <[email protected]> wrote: > > Hi all, > > I'm running a luminous cluster with tens of OSDs and > the cluster runs well. As the data grows, ceph becomes > more and more important. > > What worries me is that many services will down if the > cluster is out, for instance, the engine room is out of > electric or all ceph node are down at the same time, maybe. > > Is there a best practice on remote disaster recovery? > I wanna backup the whole ceph cluster so that the services > will come back online asap if the cluster is down. > > Thanks > _______________________________________________ > ceph-users mailing list > [email protected] > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
I would look into: - RBD mirroring if using RBD - Application level backups and remote replication (or Proxmox, OpenStack etc, whatever you are using) to be aware of data consistency - Export and import Ceph images All depends on what are the applications, what are RTO and RPO requirements, how much data, what distance, what is the network bandwidth -- Alex Gorbachev Storcium _______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
