Re: [kvm-devel] [Qemu-devel] Re: [PATCH 1/3] Refactor AIO interface to allow other AIO implementations

[Jamie Lokier <jamie@shareable.org>]

Anthony Liguori wrote:
> FWIW, in the process of optimizing the kernel driver for virtio-blk, 
> I've found that using a no-op scheduler helps a fair bit.  As long as 
> you're using a reasonably sized ring, the back-end can merge adjacent 
> requests.  This also helps a fair bit too.

By 'back-end' do you mean the virtio-blk driver itself merges
requests?

I'm assuming yes, for these questions: Does it honour barriers
properly?  Isn't that the job of the guest elevator?  Does it suggest
the various guest I/O schedulers could be enhanced to operate directly
on requests in flight in the virtio ring?

Oh, one last thing.  Is it plausible to write a Windows driver which
uses the virtio-blk interface, to get the best performing
Windows-in-KVM?  (It's nice to be able to say Linux makes a good host
for Windows guests too.)

Thanks,
- Jamie