Thank you all for your comments.
Unfortunately, IT environments are often not only about open source. Various products for various needs.
Altough we love Proxmox, it is rare that one can realistically step in an organization and say "guys, forget all your existing storage implementation and shift to CEPH in order to fit Proxmox".
ScaleIO (storage services) can easily be installed on many platforms, in mixed environments, where even Windows resources can contribute (yes, ScaleIO also installs on Windows too). The server part is not really the issue here.
EMC provides a client driver (plugin called SDC) for hosts to consume ScaleIO distributed storage and it is very easy to install it on Linux hosts. It would "just" (I know...) be a matter of integrating it to the Proxmox interface in order to make make the cluster(s) aware of the available ScaleIO resources.
In Datacenter -> Storage -> Add, we have the choice between LVM, NFS, ZFS, iSCSI etc... Are these plugins usually made available by contributors or by Proxmox developers ?
We have tried Gluster on Proxmox a while back but stumbled on some considerable stability issues and did not have time then to pursue further. However we still believe it deserve serious consideration and will be looking into it again soon.
From our experience, ScaleIO installation and use are a breeze, performance is awesome and very importantly, monitoring out of the box is feature rich and very deep, which cannot be said in aggregate for most open source storage solutions found natively on Proxmox, thus limiting for a lot of use cases.
Another two cents on transport: the prime bottleneck is often the networking for storage and RMDA (iSER / SRPt / NFS over RDMA etc.) support and awareness in the management console would improve peformance and latency tremendously while reducing CPU load significantly. Not a trivial issue in hyperconverged environments; but this is another debate...