Unfortunately, the ideal scheduler really depends on storage configuration. Gluster, ZFS, iSCSI, FC, and NFS don't align on a single "best" configuration (to say nothing of direct LUNs on guests), then there's workload considerations.
>
> The scale team is aiming for a balanced "default" policy rather than one which is best for a specific environment.
>
> That said, I'm optimistic that the results will let us give better recommendations if your workload/storage benefits from a different scheduler
I completely disagree !
If you use anything other than noop/none (depending if multiqueue is on), your scheduler inside the VM will reorder and delay your I/O.
Then the I/O will be received by the Host and this repeats again.
I can point to SuSe and Red Hat knowledge base where both vendors highly recommend noop/none as schedulers for VM.
It has nothing in common with the backend - that's in control of the hosts I/O scheduler.
Can some one tell me under which section should I open a bug ? Bugzilla is not newbie-friendly and I should admit that opening bugs for RHEL/CentOS is far easier.
The best bug section might be ovirt appliance - related , as this is only valid for VMs and not bare-metal Engine.
Best Regards,
Strahil Nikolov