Re: [PATCH v4 0/3] virtio-scsi driver

[Paolo Bonzini <pbonzini@redhat.com>]

On 02/01/2012 08:31 AM, Pekka Enberg wrote:
> What's the benefit of virtio-scsi over virtio-blk?

Most of this is in the spec or the KVM Forum 2011 presentation.

1) scalability limitations: virtio-blk-over-PCI puts a strong upper
limit on the number of devices that can be added to a guest. Common
configurations have a limit of ~30 devices. While this can be worked
around by implementing a PCI-to-PCI bridge, or by using multifunction
virtio-blk devices, these solutions either have not been implemented
yet, or introduce management restrictions.

2) limited flexibility: virtio-blk does not support all possible storage
scenarios.  For example, persistent reservations require you to pass a
whole LUN to the guest, they do not work with images.  In principle,
virtio-scsi provides anything that the underlying SCSI target supports.
The SCSI target can also be the in-kernel LIO target, which can
talk to virio-scsi via vhost.

3) limited extensibility: over the time, many features have been added
to virtio-blk. Each such change requires modifications to the virtio
specification, to the guest drivers, and to the device model in the
host. The virtio-scsi spec has been written to follow SAM conventions,
and exposing new features to the guest will only require changes to the
host's SCSI target implementation.

> Are we going to support both or eventually phase out virtio-blk?

Certainly older guests will have no virtio-scsi support, so it's going
to stay with us for a long time.

> Have the virtio specification changes been reviewed? Can we guarantee
> stable ABI for the virtio-scsi driver?

Of course.  I would have proposed it for staging otherwise.

Paolo