Hi ymechti,

in addition to Simon's answer, let me emphasize that "Openstack" is no hypervisor per se - but SUSE Openstack Cloud comes with SUSE Linux Enterprise Server, which supports KVM and Xen right out of the box.

So there are (at least) two possible migration paths, based on your description:

- drop VMware, switch to KVM or Xen, managed by Openstack - all this is combined in the product SUSE Openstack Cloud. And you'll need something to deal with the storage issue.

- keep VMware as a hypervisor, use SUSE Openstack Cloud to manage the workload and software-defined environment. Your existing storage environment may be something you can keep, if it works for you.

In the latter case, SUSE Openstack Cloud will treat your VMware servers as a single server (as it will actually interface with vCenter).

On the "RAID5 of all server disks" requirement: Of course there's technology available to run both the hypervisors and some networked storage solution on the same server machine. That would let you create sorts of a "RAID5", created from all the servers' hard disks. But I strongly advise against such a solution: One reason would be the strain this would put on both the servers I/O and the network, and secondly because I "smell" a mental approach that simply won't work with the paradigms behind OpenStack and i.e. Ceph. Been there myself, had to learn the hard way.

And more important, to me it looks like you need general design support for such a new environment. There's plenty of sizing and defining your target infrastructure that'll severely influence how you're going to spread the various workloads of an OpenStack / Ceph solution across servers. For a starter, if this is production-level and you're going to employ Neutron's SDN features (and you'll want that, if not now, then later), you'll better opt for at least two control nodes (those servers that do not carry storage nor compute workload) for both redundancy and throughput.

Deciding to change to OpenStack is easy. Designing a proper production environment is the hard part

Regards,
J