libvirt users - Sep 2016 - How to set QEMU qcow2 l2-cache-size using libvirt xml?

QEMU's default qcow2 L2 cache size is too small for large images (and small
cluster sizes), resulting in very bad performance.

https://blogs.igalia.com/berto/2015/12/17/improving-disk-io-performance-in-qemu-2-5-with-the-qcow2-l2-cache/
shows huge performance hit for a 20GB qcow2 with default 64kB cluster size:

L2 Cache, MiB   Average IOPS
1 (default)             5100
1.5                     7300
2                      12700
2.5                    63600

The above link also gives the formula:
optimal L2 cache size = L2 table size = (8 Byte) * (disk size) / (cluster size)

and the QEMU command line for setting L2 cache size, which is not persistent,
but must be specified at each invocation:
qemu-system-x86_64 -drive file=hd.qcow2,l2-cache-size=2621440

I'd like to set the l2-cache-size in libvirt xml, but cannot make out how to
do it.
There are only hints in the comments at the above link and at
http://wiki.libvirt.org/page/QEMUSwitchToLibvirt

and an earlier unanswered question on this list
https://www.redhat.com/archives/libvirt-users/2016-February/msg00036.html

Has anyone successfully done this, and could you please give an example of the
actual libvirt xml used?

Thanks!
Frank

On Mon, Sep 19, 2016 at 11:32:26AM -0400, Frank Myhr
wrote:
>QEMU's default qcow2 L2 cache size is too small for large images (and
small cluster sizes), resulting in very bad performance.
>
>https://blogs.igalia.com/berto/2015/12/17/improving-disk-io-performance-in-qemu-2-5-with-the-qcow2-l2-cache/
>shows huge performance hit for a 20GB qcow2 with default 64kB cluster size:
>
>L2 Cache, MiB   Average IOPS
>1 (default)             5100
>1.5                     7300
>2                      12700
>2.5                    63600
>
>The above link also gives the formula:
>optimal L2 cache size = L2 table size = (8 Byte) * (disk size) / (cluster
size)
>
>and the QEMU command line for setting L2 cache size, which is not
persistent, but must be specified at each invocation:
>qemu-system-x86_64 -drive file=hd.qcow2,l2-cache-size=2621440
>
>I'd like to set the l2-cache-size in libvirt xml, but cannot make out
how to do it.
>There are only hints in the comments at the above link and at
>http://wiki.libvirt.org/page/QEMUSwitchToLibvirt
>
>and an earlier unanswered question on this list
>https://www.redhat.com/archives/libvirt-users/2016-February/msg00036.html
>
>Has anyone successfully done this, and could you please give an example of
the actual libvirt xml used?
>
If you don't see it in the docs [1] it's not supported.  It would be
trivial to add that option, though.  Patches are welcome ;) If you don't
feel like it, you can always file an upstream bugzilla request [2].

Martin

[1] https://libvirt.org/formatdomain.html
[2]
https://bugzilla.redhat.com/enter_bug.cgi?product=Virtualization%20Tools&component=libvirt
>Thanks!
>Frank
>
>_______________________________________________
>libvirt-users mailing list
>libvirt-users@redhat.com
>https://www.redhat.com/mailman/listinfo/libvirt-users

Martin,

Thank you for your quick answer. I've opened a bug as you suggested:
https://bugzilla.redhat.com/show_bug.cgi?id=1377735

I'm unable to contribute a patch at the moment, but may in the future.

Thanks,
Frank