I think you are perhaps overthinking a tad. Glusterfs is a fine solution but it has had a rocky road. It would not be my first suggestion if you are seeking high level write performance although that has been improving and can be fine tuned. Instability at least in the past was mostly centered around cluster upgrades. Untouched gluster is solid and practically takes care of itself. There are definitely more eggs in one basket when dealing with hyperconverged in general. 

AFAIK ovirt does not support drbd storage nor ceph although I think ceph May be planned in true future. I’m not aware of any plans to abandon glusterfs. 

The best piece of advice I could offer from experience running hci over the past few years is to not rush to updating to the latest release right away.

On Mon, Aug 17, 2020 at 8:39 PM David White via Users <users@ovirt.org> wrote:
Hi,
I started an email thread a couple months ago, and felt like I got some great feedback and suggestions on how to best setup an oVirt cluster. Thanks for your responses thus far.
My goal is to take a total of 3-4 servers that I can use for both the storage and the virtualization, and I want both to be highly available. 

You guys told me about oVirt Hyperconverged with Gluster, and that seemed like a great option. However, I'm concerned that this may not actually be the best approach. I've spoken with multiple people at Red Hat who I have a relationship with (outside of the context of the project I'm working on here), and all of them have indicated to me that Gluster is being deprecated, and that most of the engineering focus these days is on Ceph. I was also told by a Solutions Architect who has extensive experience with RHV that the hyperconverged clusters he used to build would always give him problems.

Does oVirt support DRBD or Ceph storage? From what I can find, I think that the answer to both of those is, sadly, no.

So now I'm thinking about switching gears, and going with iSCSI instead. 
But I'm still trying to think about the best way to replicate the storage, and possibly use multipathing so that it will be HA for the VMs that rely on it.

Has anyone else experienced problems with the Gluster hyperconverged solution?
Am I overthinking this whole thing, and am I being too paranoid?
Is it possible to setup some sort of software-RAID with multiple iSCSI targets? 

As an aside, I now have a machine that I was planning to begin doing some testing and practicing with.
Previous to my conversations with the folks at Red Hat, I was planning on doing some initial testing and config with this server before purchasing another 2-3 servers to build the hyperconverged cluster.


Sent with ProtonMail Secure Email.

_______________________________________________

Users mailing list -- users@ovirt.org

To unsubscribe send an email to users-leave@ovirt.org

Privacy Statement: https://www.ovirt.org/privacy-policy.html

oVirt Code of Conduct: https://www.ovirt.org/community/about/community-guidelines/

List Archives: https://lists.ovirt.org/archives/list/users@ovirt.org/message/MHYBWFGV74OUGQJVBNPK3D4HM2FQPMYC/