There's a lot of misinformation flying around. While I suspect I know the source I'll withhold speculation and repeat the answer to the question that I've been asked 100 times over the last 3 days.
"Will Red Hat support RHEL guests running on XYZ OpenStack."
The question isn't really about supporting a RHEL guest on OpenStack it's about supporting a RHEL guest on a specific hypervisor that is being managed by OpenStack.
What really matters in this case is the hypervisor. Red Hat certifies RHEL guests running on bare metal hardware, cloud platforms and hypervisors. The management platform used to configure and control the hypervisor is not relevant to the hypervisor certification.
For example if a customer deploys RHEL running on VMware ESXi they could use VMware vCenter, CloudForms, OpenStack, vCloud Director or even command line tools to manage their environment. Regardless of the management tier the guest is certified since it is running on a certified hypervisor.
Red Hat works with industry leading cloud and hypervisor vendors including Amazon, Google, Microsoft and VMware to ensure that RHEL guests work well on their platforms. With each of these vendors Red Hat establishes engineering, QA and support relationships to ensure that the stringent enterprise requirements of our customers can be met with the appropriate SLAs.
What happens if a customer deploys RHEL on an un-certified hypervisor and encounters an issue and calls Red Hat?
Quoting from the kbase article
"Red Hat will provide commercially reasonable support efforts and refer you to the third-party software or hardware/hypervisor vendor if required"https://access.redhat.com/site/articles/1067
Note the "if required" part - we don't hang up the phone, we work with the customer to support them.
So how exactly is that lock in? How are we restricting Open Stack deployments on Mirantis, HP, PIston, Ubuntu, etc?
We're not - but let's not let little things like facts get in the way of throwing fud!
So let me turn this question around.
How would the folks at Mirantis, HP, Canonical and Piston handle a support issue with the hypervisor.
Let's say you're deploying on Mirantis and Piston and there's a problem that's diagnosed as being with the hypervisor layer - how would they handle that? In fact if the issue comes from anything below OpenStack services how would they support this? Neither company have any investment in Linux, KVM, libvirt, etc. So they have to hope that someone else in the community fixes the issue.
Considering Mirantis typically deploys on unsupported CentOS or unpaid Ubuntu I'm curious as to how support SLAs will be fulfilled in those cases.
"sorry, it's an upstream problem" isn't a response that going to cut it when a customer has an issue.
The situation is slightly
better with Canonical who do at least have some engagement in the kernel and KVM community, sadly it's so small a commitment that their contributions typically get rounded down to 0% which again begs the question what would they do when a support issue comes in?
As for HP - it's too early to say. HP only just released their HP Linux distribution inside Helion. Hopefully we'll see them ramping up their upstream investment now in Linux, KVM, libvirt, etc.