[ovirt-users] Re: Xcp-ng, first impressions as an oVirt HCI alternative
On Tue, Feb 15, 2022 at 8:50 PM Thomas Hoberg <thomas(a)hoberg.net> wrote:
Am I pessimistic about the future of oVirt? Quite honestely, yes.
Do I want it to fail? Absolutely not! In fact I wanted it to be a viable and reliable
product and live up to its motto "designed to manage your entire enterprise
infrastructure".
It turned out to be very mixed: It has bugs, I consider rather debilitating. It has also
survived critical failures in hardware, that would have resulted in data loss and
didn't because Gluster provided replicas that survived.
I have reported on both success and failure. You look through my posts, you will find
them both.
My impression is that leaders in the oVirt community have not been transparent enough
about the quality of the support they provide to the various parts. E.g. it was only very
recently that Nir wrote, that HCI was only ever supported by Gluster contributors and not
tested by the oVirt core teams.
For quite some time, ovirt-system-tests did test also HCI, routinely.
Admittedly, this flow never had the breadth of the "plain" (separate
storage) flows.
oVirt and Gluster teams did work in cooperation.
I am not sure that Nir's statement is actually that significant. It
clarifies more the organizational structure within Red Hat than the
intended quality of all relevant projects (and products). You can see
Red Hat's lifecycle pages for the relevant products to see Red Hat's
plans for them.
I think it was already clear, but if not, clarifying: From Red Hat's
POV, the replacement of Gluster is Ceph, and the replacement of oVirt
is kubevirt/OpenShift Virtualization/OKD Virtualization. This
definitely does not mean that oVirt is intended to die - on the
contrary - quite a lot of what we did in recent months was in order to
make it easier to allow oVirt to survive after Red Hat's involvement
diminishes.
I believe that the EOL of the downstream products will accelerate the dwindling of the
communty Sandro has described in his post. I honestly want to be wrong, after all I am
losing years of work and expertise.
So I posted this report on Xcp-ng, because it may be an options, who like me cannot
operate on hope alone, but need to provide a service their users can trust to have a
future without an EOL already formulated.
I would recommend that you all do a proper assessment of both platforms and potentially
others out there and learn from each other.
With factionism the state of the art cannot progress.
Agreed.
Best regards,
--
Didi