I was noting that there are other access modalities. CephFS and RGW obviate a client-side filesystem. I would think ZFS suits the OP, allowing sophisticated data protection, compression, and straightforward NFS export.
On Mar 28, 2025, at 7:48 PM, Tim Holloway <timh@mousetech.com> wrote:
Good point, although if I read that properly, all of those present the data as a raw block device and therefore have to have a client-side filesystem mounted on them. Probably more sophisticated than what our intrepid adventurer was looking for
RBD as a base image for a VM is great for cloud server VMs, though!
On 3/28/25 16:53, Anthony D'Atri wrote:
On Mar 28, 2025, at 4:38 PM, Tim Holloway <timh@mousetech.com> wrote:
We're glad to have been of help.
There is no One Size Fits All solution. For you, it seems that speed is more important than high availability. For me, it's HA+redundancy.
Ceph has 3 ways to deliver data to remote clients:
1. As a direct ceph mount on the client. From experience, this is a pain when the clients hibernate.
2. As an internal ganesha NFS server running under ceph
3. As an independent ganesha NFS server using ceph as a backend. Some deployments also do KRBD mounts onto a VM or BM system that re-exports them via KNFS
CephFS and RGW (with caveats) can also be exported with SMB.
RBD mounts, either through KRBD or libvirt are popular as well.
ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io