On Thu, Jan 24, 2013 at 2:08 PM, Anup Patel anup.patel@linaro.org wrote:
On 24 January 2013 14:55, Stefan Hajnoczi stefanha@gmail.com wrote:
On Thu, Jan 24, 2013 at 11:40:24AM +0530, Anup Patel wrote:
IMHO, If we have something like Virtio-desktop specification then all possible guest OSes can have support for it and different hypervisor can emulate it without worrying about guest support.
At this point x86 virtualization is mature and working with a mix of emulated x86 architecture pieces and virtio devices for performance-critical or open-ended functionality that we want to be able to extend.
ARM is getting KVM and virtio-mmio support. It will be in a similar position soon.
Virtio guest drivers have not been implemented widely. The Linux and Windows efforts are driven by the folks who were behind virtio from the start, but Solaris, FreeBSD, and others didn't really jump on the virtio bandwagon.
[Anup] I think other OSes will be motivated to added Virtio drivers if there exists some think like Virtio-desktop specification that is being emulated by many hypervisors.
Absolutely, if most hypervisors implement it then guests will also implement it. But this is exactly what I tried to describe in the previous email:
Yes, virtio has the potential to be implemented by many hypervisors. The specification is out there and existing open source implementations are out there.
But the leading x86 hypervisors (kvm, xen, vmware, hyperv) implement their own paravirt I/O approaches. This may be due to historical reasons but virtio has been around long enough for any of the other big hypervisors to implement it if they wanted to.
Code sharing or a unified standard isn't sufficient motivation when there are other factors like ability to change device spec to achieve better performance, compatibility with existing guests, ability to add new features, certified device drivers, etc.
The incentives need to be favorable for hypervisors to wed themselves to a standardized paravirt device - simply adding more virtio devices doesn't change today's incentives, so you cannot expect hypervisor vendors to act differently from how they have in the past.
Stefan