Ceph is designed to handle reliability in its system rather than in an external one. You could set it up to use that storage and not do its own replication, but then you lose availability if the OSD process hosts disappear, etc. And the filesystem (which I guess is the part you're interested in) is the least stable component of the overall system. Maybe if you could describe more about what you think the system stack will look like? -Greg Software Engineer #42 @ http://inktank.com | http://ceph.com
On Tue, Apr 8, 2014 at 2:38 PM, Franks Andy (RLZ) IT Systems Engineer <andy.fra...@sath.nhs.uk> wrote: > Hi All, > > First post, so please excuse any ignorance! > > We have a need to provide HA storage to a few thousand users, replacing our > aging windows storage server. > > I would like to use Ceph alongside (being sensible) xfs - or zfs if I'm > feeling game, but need to understand something critical. Ceph it seems > hinges on multi server provided storage, i.e. building a collection of > protected storage from multiple hosts and managing those hosts' underlying > physical or virtual storage. Our storage all comes from a group of > equallogic SANs, and since we've invested in these and vmware, the obvious > choice for any hosts going forward are virtual, with iscsi backend storage > provision, or maybe vmfs provided partitions (not preferred over direct > iscsi). > > Our storage cluster mentioned above needs to export SMB and maybe NFS, using > samba CTDB and whatever NFS needs (not looked into that yet). My question is > how to present the storage ceph needs given that I'd like the SAN itself to > provide the resilience through it's replication and snapshot capabilities, > but for ceph to provide the logical HA (active/active if possible). > > How does it work with single shared backend iscsi targets, or am I going > down the wrong path altogether? > > I could just set up a load balancer to point at a couple of hosts with the > same single iscsi target mounted. Samba would take care of the file locks, > but any snapshotting or similar obviously need to quiesce the volume, which > might cause all sorts of nastiness in an active/active cluster. I would want > the active/active bit to be at a file level obviously - multiple connections > to the same individual files just wouldn't work. > > Any pointers? Anyone done this sort of thing before? > > Perhaps I'm better off creating a replicated couple of volumes and using > ceph to manage those, providing active/active alongside replication? Or > should I stick to active/passive? There's no great need for active/active > but it seems a waste to have a redundant host doing nothing unless it's > absolutely necessary. > > I've tried a couple of general linux forums but I'm either asking the wrong > questions, it's stupidly obvious or people just don't know! > > Thanks > > Andy > > > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
