Future of RHV / Veeam for RHV / OpenShift

[mdws]

Hello,

Ovirt 4.5 was released an solves some of the problems with the backups (VM locking, fixed size scratch disk...).
Does anyone know if there are plans from RedHat / IBM to ever release a version 4.5?
In the Roadmap there is no 4.5:
https://access.redhat.com/support/policy/updates/rhev
Will Veeam for RHV ever be released if there is no RHV 4.5?

[HannesK]

Hello,
I can only speak for Veeam: we have plans for a future for the product.

Best regards,
Hannes

[mdws]

Hello,

we have now the Statement from our Software Dsitribution that there will be no RHV 4.5 and RHV will be not orderable anymore after Q3.
The successor is Red Hat OpenShift Virtualization https://www.redhat.com/en/technologies/ ... ualization
What are Veeams Plans? Support Ovirt 4.5? Support Red Hat OpenShift Virtualization in the Future? No support for KVM?

[cody.ault]

RHV 4.4SP1 which was recently released is based on oVirt 4.5.

[mschau]

Hello,

any news on this in the meantime?
Any plans for OpenShift Virtualization?

[HannesK]

Hello,
there are no news to share in public for now. As soon as we see, where RHEV / clones are are shifting, that might change.

As for OpenShift Virtualization: once it has proper change block tracking, maturity and market adoption, we will look at it.

Veeam will only support hypervisors with proper change block tracking (CBT). We don't want to send our customers back to the "middle ages" of backup.

We have alternatives for any hypervisor: agent-based backup. The agents for Windows and Linux have change block tracking built in.

Just curious: are you migrating to OpenShift virtualization (how many machines)? I recently talked to some Red Hat guys and they were not really positioning that as alternative to RHEV.

Best regards,
Hannes

[cody.ault]

I would add the OpenShift Virtualization has basic protection built-in to our Kubernetes offering K10 (Kasten)
https://docs.kasten.io/latest/usage/ope ... ualization

[criani]

Situation - We are currently using Linux agent on a handful of VM's. The VM's in question are KVM and are used for product development. As such, we will often unmount and modify LVM's and remount (or mount a new LVM). This is not going well with the agent and is making managing VM's a bit of a hassle.
Best Laid Plans - We were excited to see that Veeam supported an integration with RHV, which would have allowed us to easily managed from the host level, making managing our backups easier. Unfortunately, while RHV is still supported, the license is not longer being sold. We have been told that RedHat OpenShift is the new hotness, and the comparable license to RHV (RHV capabilities are seemingly a small part of the overall Openshift license).

The actual question - Is Veeam currently (or planning to) integrate with RedHat OpenShift?

[Gostev]

While RedHat may think OpenShift is the new hotness, we're not necessarily convinced at this time. As such, there are no specific plans to jump on this right away. But we are definitely keeping an eye on its adoption and will act accordingly. We never say never but there has to be a significant market opportunity before we invest our R&D resources... we don't usually do things just because they are hot.

[criani]

Gostev, let me clarify, it's not just the new hotness, RedHat is no longer selling licenses for RHV, we have been told Openshift is the only option.

[cody.ault]

Veeam K10 has some basic protection for these workloads today that include full/incremental backups and restores:
https://docs.kasten.io/latest/usage/ope ... ualization
Some of the main features that are currently missing between a standard virtualization platform and Openshift Virtualization are things like Changed Block tracking and File Level Recovery. Since you mentioned these are dev systems, can you elaborate on what types of restore you do most often and features you think might be missing?

[Gostev]

Gostev, let me clarify, it's not just the new hotness, RedHat is no longer selling licenses for RHV, we have been told Openshift is the only option.

"Sacred place is never empty" as they say. I have heard of at least two major vendors looking to pick up the ball dropped by Red Hat.

[mdws]

95% of our customers are running their VMs on Enterprise SAS / FC Storages and they are only runnng VMs, no containers, on the Hypervisor.
When they run containers, they run it on VMs.
Openshift / Kubevirt will AFAIK never support shared Block Storage with Snapshots / CBT an so no one will take a look at it when they plan to replace vSphere.

As many VMware customer plan to switch to another Hypervisor now, would it not make sense for Veeam to pick up the RVH / oVirt ball?
oVirt is a 1:1 replacement for vSphere when we talk about classic virtual Machine operating and has some advantage over Proxmox, for example:

- shared and working Block Storage with FC / iSCSI Enterprise Storages
- Snapshots and CBT for VMs on shared Block Storages

Proxmox has still no option to use a Enterprise SAN and I don't expect that they will change this, it's too complex to handle LVMs with Snapshots, CBT and automatic expansion compared with NFS / CEPH, in oVirt this is handled by the SPM which is a unique oVirt feature, until now.

When I offer my customers a CEPH / NFS hyperconverged solution they shake their heads as they will never replace a robust high performance solution with a cheap but slow SDS solution.

[ashleyw]

5% of our customers are running their VMs on Enterprise SAS / FC Storages and they are only runnng VMs, no containers, on the Hypervisor.
When they run containers, they run it on VMs.
Openshift / Kubevirt will AFAIK never support shared Block Storage with Snapshots / CBT an so no one will take a look at it when they plan to replace vSphere.

Has anything changed in this space as this is currently our exact scenario and we've been trying to figure out an exit strategy from VMware?

We have evaluated Proxmox, and we hit the same issue of no native block storage support with FC, without presenting FC LUNs to individual nodes and then using SDS like Cepth (or similar) to stitch the storage together - but this is a far cry from a VMFS presented FC LUNs - so for our cluster tests we had to use NFS which is unlikely to scale at high volume(say 400-500 VMs).

After RedHat recently announced a new virtualisation focused OpenShift product (Presumably leveraging the KubeVirt layer in Openshift and providing a roadmap to switch to container based workloads over time), I stood up a 3 node OpenShift Cluster last night.
However, without switching to a high performance NFS system, the same shared storage dilemma exists (I think).
The Veeam K10 product does not appear to be integrated into the existing Veeam management console we know and love.
Our existing allflash FC SAN (Unity 680F) is centered around delivering block based storage over FC. Even though it has dual 4x25GB IO cards in and can technically handle NFS, the view from our teams is that this is not scalable to same same performance as FC block storage.

Switching to agent based VM backups would slow our backup throughput dramatically.
Switching from VMware and still retaining FC based SANs and getting the same level of functionality/performance/reliability appears to be very difficult.
I Just want to make sure I haven't overlooked anything here...

[ashleyw]

Also I guess NFS doesn't support storage snapshots so we'd take a big hit on performance if we went down that road.

[Gostev]

Assuming the question is "future of RHV", then from what I recently heard from a partner, Oracle really picked up the oVirt flag and are apparently working on the following at the moment:
1/ Replacing old RHEL with a fresher Linux modern kernel version
2/ Collaborating with Lightbits to add support for NMVe-OF