David Brown <david@westcontrol.removethisbit.com> writes:
>Judging from the history (with SCSI-3 in 1996, SAS in 2001, and USB 2 in
>2000) I would have thought that parallel SCSI-2 would be well out of
>date before USB disk adaptors started becoming popular.
Just because the spec was out, doesn't mean everything instantly cut over.
I was buying parallel SCSI based servers well into 2006-2007 timeframe.
>relatively low number of parallel SCSI (rather than SAS) installations
I'd have to say that parallel SCSI installations far outnumbered SAS
given how many years it was available.
>Of course, if you know of such devices, then I'll just have to be surprised!
> David Brown<david.brown@removethis.hesbynett.no> writes:
>> On 23/05/12 18:54, Robert Lacoste wrote:
>>> Dear all,
>>> We are looking for an USB to SCSI-2 bus interface chipset. Not a
>>> ready-made interface cable, but either a dedicated bridge chip (I'm
>>> afraid they no longer exist...) or a microcontroler/FPGA with the
>>> associated firmware SCSI stack. Any available reference design or
>>> commercial solution somewhere, or should we start from scratch, which is
>>> of course possible but quite long ? Custom host-side drivers would not
>>> be a huge problem in that case.
>
>> That's an odd requirement - parallel SCSI was never particularly
>> popular, and SCSI-2 was outdated before USB 2.0 came on the scene (if
>> I've got my history right). I doubt if a dedicated bridge chip ever
>> existed for something like that.
>
> ?? Parallel SCSI never popular? Where were you? It was huge in the PC
> server, workstation (back when they were separate than desktops), and
> RISC server markets, with almost total penetration?
>
I meant relative to the alternative - IDE/ATA - parallel SCSI was of
limited popularity. It had, as you say, almost total penetration in
particular markets for a while, but it was still only a tiny percentage
of the total disk market.
> There are many commercial USB to SCSI bridges you can buy.
Judging from the history (with SCSI-3 in 1996, SAS in 2001, and USB 2 in
2000) I would have thought that parallel SCSI-2 would be well out of
date before USB disk adaptors started becoming popular. Based on the
relatively low number of parallel SCSI (rather than SAS) installations
still in use since USB 2 became mainstream, and the low percentage of
those that would be interested in a USB-SCSI adaptor (think about the
use-cases - they are much less than for USB-IDE or USB-SATA), I find it
surprising to think that it would be worth making such a bridge chip at all.
Of course, if you know of such devices, then I'll just have to be surprised!
> Unfortunately, they are all old designs, based on chips that have
> long since been end-of-lifed. (ie. I think Prolific had one. Maybe
> Oxford Semi, etc).
>
> The OP is looking for a still current maker, which may be quite a no-go.
>
Reply by Robert Lacoste●May 24, 20122012-05-24
"Doug McIntyre" a �crit dans le message de groupe de discussion :
4fbd6e9d$0$74663$8046368a@newsreader.iphouse.net...
>The OP is looking for a still current maker, which may be quite a no-go.
That's exactly the problem... I guess that there is no longer any active
USB-to-SCSI2 chip available, but we were hoping that someone has already
developped that on an FPGA or ARM micro, to replace such a chipset. There
are still plenty of old SCSI peripherals that some customers want to keep
using...
Robert
Reply by Doug McIntyre●May 23, 20122012-05-23
David Brown <david.brown@removethis.hesbynett.no> writes:
>On 23/05/12 18:54, Robert Lacoste wrote:
>> Dear all,
>> We are looking for an USB to SCSI-2 bus interface chipset. Not a
>> ready-made interface cable, but either a dedicated bridge chip (I'm
>> afraid they no longer exist...) or a microcontroler/FPGA with the
>> associated firmware SCSI stack. Any available reference design or
>> commercial solution somewhere, or should we start from scratch, which is
>> of course possible but quite long ? Custom host-side drivers would not
>> be a huge problem in that case.
>That's an odd requirement - parallel SCSI was never particularly
>popular, and SCSI-2 was outdated before USB 2.0 came on the scene (if
>I've got my history right). I doubt if a dedicated bridge chip ever
>existed for something like that.
?? Parallel SCSI never popular? Where were you? It was huge in the PC
server, workstation (back when they were separate than desktops), and
RISC server markets, with almost total penetration?
There are many commercial USB to SCSI bridges you can buy.
Unfortunately, they are all old designs, based on chips that have
long since been end-of-lifed. (ie. I think Prolific had one. Maybe
Oxford Semi, etc).
The OP is looking for a still current maker, which may be quite a no-go.
Reply by David Brown●May 23, 20122012-05-23
On 23/05/12 18:54, Robert Lacoste wrote:
> Dear all,
> We are looking for an USB to SCSI-2 bus interface chipset. Not a
> ready-made interface cable, but either a dedicated bridge chip (I'm
> afraid they no longer exist...) or a microcontroler/FPGA with the
> associated firmware SCSI stack. Any available reference design or
> commercial solution somewhere, or should we start from scratch, which is
> of course possible but quite long ? Custom host-side drivers would not
> be a huge problem in that case.
> Thanks,
> Robert
> www.alciom.com
>
>
That's an odd requirement - parallel SCSI was never particularly
popular, and SCSI-2 was outdated before USB 2.0 came on the scene (if
I've got my history right). I doubt if a dedicated bridge chip ever
existed for something like that.
Are you looking for a one-off solution? I would think the realistic
solution is to put together an embedded PC card with a SCSI interface
card, running Linux, and pass the disk on as an iSCSI disk on the
network rather than USB.
Reply by Robert Lacoste●May 23, 20122012-05-23
Dear all,
We are looking for an USB to SCSI-2 bus interface chipset. Not a ready-made
interface cable, but either a dedicated bridge chip (I'm afraid they no
longer exist...) or a microcontroler/FPGA with the associated firmware SCSI
stack. Any available reference design or commercial solution somewhere, or
should we start from scratch, which is of course possible but quite long ?
Custom host-side drivers would not be a huge problem in that case.
Thanks,
Robert
www.alciom.com