KVM Virtio-blk x-data-plane support


I wonder if KVM Virtio-blk’s new x-data-plane feature support by Cloudmin UI will take place or considered any how? i would like to report I have tested Virtio-blk with x-data-plane = on Host machine Ubuntu 14.04 an ibm x3850 with SSD drives with a great performance. it must be xdefault mode definitely for Virtio block devices. here is a document you may want to see. http://events.linuxfoundation.org/sites/events/files/slides/CloudOpen2013_Khoa_Huynh_v3.pdf

but the things is i had to edit guest settings file manually to achieve this. result was ok but Cloudmin UI since i had to change kvm startup parameters in a way that causes Cloudmin can not see disk device in list. “if=none” parameter makes it fail to parse kvm parameter line. I’m not a professional Perl programmer. can’t tell more.

I would like to help for Cloudmin support for Virtio optimizations if possible more like code contribution.

Sounds like a good idea. I’ll file a ticket in the tracker, so Jamie can look into it. Sounds easy enough to support.

Ticket is here: https://www.virtualmin.com/node/35167

So are there any downsides or limitations to enabling x-data-plane=on ? I’m wondering if Cloudmin should just turn it on by default where supported…

Actually, I found some docs at http://blog.vmsplice.net/2013/03/new-in-qemu-14-high-performance-virtio.html that do list downsides :

  • Image formats are not supported (qcow2, qed, etc).
  • Live migration is not supported.
  • QEMU I/O throttling is not supported but cgroups blk-io controller can be used.
  • Only the default “report” I/O error policy is supported (-drive werror=,rerror=).
  • Hot unplug is not supported.
  • Block jobs (block-stream, drive-mirror, block-commit) are not supported.

None of those seem like deal-breakers though.

Well i knew the downsides but as you say they don’t seem like deal-breakers.
Anyway maybe not on by default. but can be a new option right?