On Fri, Sep 30, 2022 at 7:36 PM Filipe Mendes <filipehdbr@xxxxxxxxx> wrote:
>
> Hello!
>
>
> I'm considering switching my current storage solution to CEPH. Today we use
> iscsi as a communication protocol and we use several different hypervisors:
> VMware, hyper-v, xcp-ng, etc.
Hi Filipe,
Ceph's main hypervisor target has always been QEMU/KVM but xcp-ng
(i.e. Xen) has native support support as well.
Starting with pacific, RBD is natively supported on Windows. Lucian
and the team are continuously improving it so Hyper-V use case should
be covered as well.
>
>
> I was reading that the current version of CEPH has discontinued iscsi
> support in favor of RBD or Nvmeof. I imagine there are thousands of
> projects in production using different hypervisors connecting to ceph via
> iscsi, so I was curious that I did not find much discussion on the topic in
> forums or mailings, since so many projects depend on both: ceph + iscsi,
> and that RBD only communicates well with Proxmox or openstack. Also nvmeof
> is not fully supported on ceph and many other popular hypervisors.
Upstream, ceph-iscsi gateway isn't going anywhere -- at least not
until ceph-nvmeof gateway matures enough (currently it's very much
WIP). I don't expect any active development to happen on ceph-iscsi
but it should still be maintained and the packages would remain
available.
>
>
> So the trend is that other hypervisors will start to support RBD over time,
> or that they will start to support nvmeof at the same time that ceph
> implements it stably?
>
>
> Am I missing or maybe mixing something?
I'm not sure what you mean by NVMeOF not being fully (?) supported on
various hypervisors. Strictly speaking you don't need any support in
the hypervisor at all: as long as your guest OS supports NVMeOF, you
can set it up in the VM itself, exactly the same as with iSCSI.
Thanks,
Ilya
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
