Subject: Re: SCSI network
To: None <>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: tech-kern
Date: 10/02/1999 17:47:40
[Replying to multiple messages here.]

> From: "Frank Warren" <>

> I think you're probably on SCSI or early SCSI 2 with fast-wide at
> most.

Definitely not wide, and I think non-fast.  The machines I'm
contemplating doing this on are SPARCstations (SLC, ELC, IPX, 1+) and
possibly a Sun-3/60.  The former use the ncr53c9x driver, if my reading
of dev/sbus/files.sbus is accurate.  What indications I find imply that
the Sun-3 uses the NCR5380.

> The real problem with SCSI is that it is not really designed for
> networking.  The address space is limited to 8 devices in SCSI, 16 in
> SCSI-2.  It won't be a big network :-)

I'll be quite satisfied to put two hosts on it. :)  (Two hosts, plus
one disk drive for one, two drives and a tape for the other, that's six
devices right there.  It's going to be a bit crowded. :)

> An interesting way of extending this network, if you can add SCSI
> controllers, would be to base them on a star topology with, say, your
> Sun 3 acting as the router.

Well, not a Sun-3/60, since one can't add SCSI controllers (or much of
anything else) to that.  Maybe one of the -3/260s, since those can have
extra SCSI boards slapped in them...and winter is coming, so the heat
problem won't be so bad.

> One problem you will DEFINITELY run into is that few SCSI controllers
> ever have responded to the SCSI IDENTIFY command.  This makes dynamic
> configuration of your network a problem.

I'd be quite satisfied - at least for the first cut - to hardwire
everything in the kernel config.

> From: Paul A Vixie <>

> this sounds like RFC 2143, [...]

Ah!  Thanks for the pointer.  I hadn't thought much about the details
of the encapsulation - I may or may not end up following 2143, but am
glad to have something there.

> From: Matthew Jacob <>

> The actual sensible way to do this is IP encapsulation over
> FibreChannel SCSI.

That completely defeats the purpose of my doing it in the first place,
which is to get a second network interface on SLC and ELC machines
running headless (got 'em for free when the video electronics died).
One can't add diddly to the things - they have an SBus as far as
software is concerned, but no hint of it appears in terms of any of the
connectors on the thing.

> From: Matthew Jacob <>

> There *is* a protocol for negotiating parallel SCSI IDs with multiple
> initiators- it's called SCAM (SCSI Configured Automagically- I kid
> you not, it's even in various ANSI documents).

A while ago, I fetched a SCAM draft off some Seagate distribution point
and read it over.  What I recall of it makes me think it'd be
interesting to do but not really very relevant to using SCSI as a
networking medium.  Does that match what you have in mind?

					der Mouse

		     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B