virtio for 9.1-R
Constantine A. Murenin
mureninc at gmail.com
Wed Nov 28 19:53:13 UTC 2012
On 27 November 2012 13:25, Paul Mather <paul at gromit.dlib.vt.edu> wrote:
> On Nov 27, 2012, at 3:49 PM, Joe Holden <lists at rewt.org.uk> wrote:
>> On 27/11/2012 19:25, Sergey Kandaurov wrote:
>>> On 27 November 2012 22:12, Joe Holden <lists at rewt.org.uk> wrote:
>>>> Hi guys,
>>>> I can't see virtio in releng/9.1, is there any particular reason why it
>>>> isn't going to be included given that it works reasonable well (and is
>>>> optional anyway, so not likely to be detrimental)?
>>> virtio appeared in stable/9 a bit after 9.1 cut off,
>>> and it is too late now regardless of virtio shape.
>>> Anyway you can installed it from ports.
>> Ah I see, doesn't really help all the people who can't install it in KVM and such though unfortunately, seems silly making them wait even
>> longer and having to use Linux :)
> FWIW, I installed FreeBSD 9-STABLE (pre-virtio in src) in KVM, initially using the emulated devices and then, post-install, installed the emulators/virtio-kmod port and switched over to vtbd/vtnet devices without problem. When virtio appeared in src, I ditched the virtio-kmod port, again, without issues.
> I found making the transition to virtio devices no harder than the ad -> ada device transition.
What's the actual reallife performance benefits from virtio? Twofold,
threefold, more, less?
Linux-KVM web-site, http://www.linux-kvm.org/page/Virtio, mentions
that they can push 1Gbps or so through the virtio network driver, but
the number seems meaningless, since no context is established of the
KVM performance prior to virtio.
Another official-looking page, http://wiki.libvirt.org/page/Virtio,
fails to make any number-based sales pitch whatsoever, whereas
http://www.linux-kvm.org/page/Using_VirtIO_NIC likewise attempts to
show up some random meaningless numbers, still failing to establish
the context of what the performance is like without virtio.
More information about the freebsd-stable