RE: ATA/IP & ATAPI

[julian_satran@il.ibm.com]

Costa,

The points you make are very relevant. The large installed base makes a
good argument to have ATA/ATAPI devices as targets. For initiators (hosts)
having a SCSI driver or  an ATA driver makes no difference.
I don't see any major issue with having a SCSI to ATA convertor on the
target. This will cover most of the appliances, RAID or JBOD boxes.  For
new devices I think that SCSI has only a moderate level of complexity and
costs are only a mater of volume.

Regards,
Julo

Costa Sapuntzakis <csapuntz@cisco.com> on 15/02/2000 19:06:09

Please respond to Costa Sapuntzakis <csapuntz@cisco.com>

To:   "Bradley, Mark" <mark_bradley@btc.adaptec.com>
cc:   ips@ece.cmu.edu (bcc: Julian Satran/Haifa/IBM)
Subject:  RE: ATA/IP & ATAPI





Installed base is a very good argument. :) Here are some of the
issues:

Many ATA drivers are written to do register-level accesses. Rewriting
them to send packets will involve some serious reworking.

An ATA/IP device could maintain a register interface to the host with
a special ATA/ISA<->ATA/IP bridge (either emulated in BIOS or as a
real PCI device).

Currently, all ATA devices plug into motherboards with
processors. These processors could act as SCSI<->ATA converters.
Or are there ATA-only functions that are lost by that sort of bridging?
Perhaps SMART...

Would an ATA/IP device be significantly simpler than a SCSI/IP device
from a firmware standpoint?

-Costa

On Tue, 15 Feb 2000, Bradley, Mark wrote:

> About 85% of storage on IA32/64 systems is IDE/ATA.  Ignoring
> this substantial a volume seems inappropriate.  Further, there
> is a proposal for Serial ATA (SAT) that might lend better lend
> itself to this work.
>   --  markb
>
> > -----Original Message-----
> > From: Costa Sapuntzakis [mailto:csapuntz@cisco.com]
> > Sent: Tuesday, February 15, 2000 1:07 AM
> > To: ips@ece.cmu.edu
> > Subject: ATA/IP & ATAPI
> >
> >
> >
> > Hi,
> >
> > What are the arguments for ATA/IP?
> >
> > -Costa
> >
>
>
>