> > PS. SCSI over IP is VERY strange thing anyway, but why defining ATA 
> > over IP too it's complete nonsense. SCSI over IP could be ok for any 
> > media (other side don't need to know if it's  IDE or SCSI)
> 
> SCSI-over-IP (a.k.a iSCSI -- and it's actually SCSI-over-TCP) isn't any 
> more strange than SCSI-over-FC (which, of course, already exists).
> 
> And, if you're using the SATA framing protocol, ATA commands over any 
> other packet-oriented network isn't really much of a stretch.  Of 
> course, there are a lot of good reasons to just do SCSI over those 
> packet-oriented networks, since SCSI has a much richer command set and 
> architecture model.

I concur with all of that.

The original post was for ATA over *ethernet*, not IP.
I looked at the spec. It is extremely light weight.
Conceivably it could be implemented in an ASIC.

I wouldn't go so far as to say it's a good idea.
It's really just good enough for booting, and
the world has enough of those.

Sometimes I wish for a disk-over-ethernet protocol
without all the ceremony of IP, let alone TCP.
BOOTP/DHCP and TFTP usually gets me through.

Still, I can see why somebody would want to do
something like ATA-over-ethernet. It would be
very helpful early in the bring-up process. iSCSI
or NFS to fill such a requirement is a little like
using a blast furnace to light a bar-b-que.

-gww