T.R | Title | User | Personal Name | Date | Lines |
---|
2837.1 | in the last 10 or so | WJG::GUINEAU | Opening the doors of Perception | Thu Aug 17 1989 11:18 | 4 |
| Look at another very recent note in thsi conference which addresses
this very issue concerning a VAX and Amiga
John
|
2837.2 | How about X-windows? | SMAUG::SPODARYK | Sick for toys! | Thu Aug 17 1989 11:42 | 18 |
| Well, it'd be expensive, but you could buy Dale Luck's X-windows,
and the necessary communication hardware, and use that to share
resources. Use the 500 to run some X clients and display the output
on the 2500. Or vice versa.
Dale demo'ed a standard 2500 and a Unix 2500+68030 doing just that,
and it worked beautifully. The performance was exceptional. He was
running TCP/IP over Ethernet.
Right now, the choice of transports is pretty small. They are working
hard to allow a variety of protocols to run over a variety of
transports. Ie. TCP/IP, DECnet, etc - over - Ethernet, serial, etc.
The hooks are in place for DECnet, but I don't believe it's available
yet.
There are discussions of this elsewhere.
Steve
|
2837.3 | | BAGELS::BRANNON | Dave Brannon | Thu Aug 17 1989 12:06 | 11 |
| Don't forget all those games that can be played over a modem (or
in this case, a null modem).
re:.2 anybody heard if midi can be one of the transports? Seems
like it would make a great low speed network, with adapters already
available, some even with serial port pass thru.
In the ST world, Midi-Maze uses it for transport to talk to multiple
STs. Nice game, is anybody working on an Amiga version of it?
-Dave
|
2837.4 | DECnet is available. | AYOV28::ATHOMSON | C'mon, git aff! /The Kelty Clippie | Thu Aug 17 1989 12:17 | 13 |
|
re: .2
� The hooks are in place for DECnet, but I don't believe it's available
� yet.
There was a small article in the July Amiga Computing International
announcing DECnet (endnode) for the Amiga. I can remember that NCP
and NFT were supported, but mail and some others were to be in the
next release. I'll dig out the article tonight and post it tomorrow.
Alan T.
|
2837.5 | I am unknowlegeable, I have to look that stuff up | GIAMEM::I_SHAW | 13 days | Thu Aug 17 1989 12:42 | 11 |
| The previous note talks about VAX-Amiga. Wouldn't it be much easier
on an Amiga-Amiga system? I don't know about Ethernet or TCP/IP (I'll have
to look them up in here).
Is it possible to link the two (just hardware wise) without using the
serial port?
If not, is there an in-board modem you can buy for the 2xxx to bypass
the serial prt?
--mikie--
|
2837.6 | | TALLIS::MCAFEE | Steve McAfee | Thu Aug 17 1989 13:07 | 6 |
|
You might want to look back through some recent notes for something
called DNET written by Matt Dillon. I believe this works between
amigas.
-steve
|
2837.7 | Summary written at 30�C | FRAMBO::BALZER | Christian Balzer DTN:785-1029 | Thu Aug 17 1989 13:34 | 33 |
| Re: .5
There are several options:
1. Use Ethernet (running TCP/IP, DECNet, younameit). There are several
cards available, but this is going to be the most expensive approach
(~$1000 at least for both systems). But this is also going to be
very fast and "future compatible".
2. Wait for the announced ARCNet cards from CBM (A2060, A560). These
should perform a little slower than Ethernet, but will probably
be much cheaper.
3. You can use the parallel port for communication, it's
bi-directional. CBM used (still uses?) this method in their development
environment for fast data transfer. Unfortunately I can't think
of any decent program that supports this, especially DNet would
be a killer with this. However, you would loose your (parallel)
printer connection.
4. There are a number of serial/modem boards available for the A2000.
This way you could use the builtin port for communication with the
A500. Since there are also multiple serial AND parallel boards out
there, you might run a multi-user BBS on the A2000, communicating
with the A500 via the (builtin) parallel port and printing some
documents using an additional parallel port on a card.
Since the cards in #4 also cost money, #3 would be the cheapest
solution, given that you or someone else whips up the software.
Regards,
<CB>
|
2837.8 | Amiga - The Next Generation | WJG::GUINEAU | Opening the doors of Perception | Thu Aug 17 1989 14:57 | 9 |
|
How about taking that old 500 and hacking up the motherboard to plug into
an A2000 slot. A mere matter of microcode would get you a
Multi-Processor (already) Multi-Tasking Amiga (A4000?)
Any takers?
John (the instigater)
|
2837.9 | make a fiber link | ANT::JANZEN | cf. ANT::CIRCUITS,ANT::UWAVES | Thu Aug 17 1989 15:29 | 2 |
| The 68000 isn't microcodable
Tom
|
2837.10 | Where's DNET 2.??? Not on NORSE... | GIAMEM::I_SHAW | 13 days | Thu Aug 17 1989 16:00 | 9 |
| Where is DNET? I looked through previous notes and found a reference
to it on NORSE, but alas, it isn't there. Has it been recalled for selling
purposes? I think this may be the answer for me. I like the idea of the
parallel connection (add another to the 2xxx for printer) and client-type
operation.
Please help me find it! I won't get the 2xxx for a while, and I could
examine the stuff and maybe hack it for my needs...
--mikie--
|
2837.11 | | UFP::LARUE | Jeff LaRue - MAA Senior Network Consultant | Thu Aug 17 1989 16:17 | 9 |
| Depending on what you wish to do between the two
(or more) Amigas, you could use the SCSI bus.
That would make it an:
AmigaCluster!!
-Jeff
|
2837.12 | DNET on TAPE | ELWOOD::PETERS | | Thu Aug 17 1989 18:08 | 10 |
|
re .10
Dnet version 2.0 is on Fred Fish Disk 220. The disk is in
TAPE::USER1:[AMIGA.FISH.FF220]FF220.ZOO
Steve Peters
|
2837.13 | | POBOX::ANDREWS | I'm the NRA | Thu Aug 17 1989 18:18 | 6 |
| Don't forget to get the NET: device stuff from the Software Distillery.
This allows you to access devices on the other systems as if they
are local devices. I tried doing this the other day and it was
pretty neat! I ran a program from a "remote" system (about 2 feet
away) just to see if it could be done. The NET: device is built
on top of DNET.
|
2837.14 | Transactor | WJG::GUINEAU | Opening the doors of Perception | Thu Aug 17 1989 18:46 | 8 |
| There was a nice writeup on the Software Distillery's NET: device and how
it works.
Very Clever! Wonder if those guys will support UNIX machines (DNET does).
Now that would be nice!
John
|
2837.15 | WHERE's NET:??? | FRAMBO::BALZER | Christian Balzer DTN:785-1029 | Fri Aug 18 1989 05:21 | 18 |
| Re: last few
Yeah, NET: sounds VERY nice, unfortunately I can't seem to able
to get my hands on it. Would the gentleman who shared his experiences
with this software a few notes back be so kind and upload it to
TAPE::, NORSE:: WJG::, etc.
On the topic of SCSI-NET, the only implementation I know of is the
Cltd. stuff. Now, if you don't care for DMA/fast and efficient data
transfer, these controllers are for you. :-) (Read my HD controller
review some hundred notes back) Or if you have enough spare cash,
you might use the Cltd stuff for networking and get a "real"
controller for your HD(s)...
Regards,
<CB>
|
2837.16 | What "features" are wanted? | LOWLIF::DAVIS | That's not a BUG, it's a FEATURE! | Fri Aug 18 1989 11:17 | 20 |
| re: .11
>Depending on what you wish to do between the two
>(or more) Amigas, ...
That's the key. Before you can talk about "how" to do it you need to figure
out the "why" of doing it. What are you trying to accomplish - file
transfers, remote devices, etc?
>That would make it an:
> AmigaCluster!!
I like it! That way a person could get a hard disk on one system and have
a few 500's clustered in and share the drive! :-) Maybe we could even get
some distributed queing??? Yeah, and cluster-wide process control like in
VMS 5.2. Yeah, that's the ticket!
Dreaming along,
...richard
|
2837.17 | | BAGELS::BRANNON | Dave Brannon | Fri Aug 18 1989 12:10 | 14 |
| re:.16
That sharing of the harddisk is possible now with Cltd's SCSI-net.
How do we convince the other vendors that
1. it's a great idea
2. implement it in a compatible way so that it will work with
other vendor implementations of SCSI-net
But how do you get beyond the 8 node SCSI limit? (with 2 systems
with 3 imbedded SCSI drives each, you reached the limit. Hmmm..
what if you had two controllers, each going to a different SCSI
bus.... and then add routing.... naw... too silly.
-Dave
|
2837.18 | My needs defined (sorta) | 21348::BELL | | Fri Aug 18 1989 16:21 | 17 |
| Ok, I guess I'll define what I'd like to be able to do (which sounds
a lot like DNet):
Running DNet over the parallel port, I'd like to get on the 2500 and:
- Freely swap files from system to system
- Run programs (using Intuition) on the 500; i.e. use the 2500 as a
remote graphic terminal
- Start-up a remote CLI
I guess number two is the most important. To be able to have full
control over what goes on on the other system by either CLI or actual graphic
contact.
--mikie--
|
2837.19 | partial solution | WJG::GUINEAU | Opening the doors of Perception | Fri Aug 18 1989 17:19 | 7 |
|
re .16
> But how do you get beyond the 8 node SCSI limit? (with 2 systems
Logical Units
|
2837.20 | I need NET: too!!! | BSS::BRANDT | | Fri Aug 18 1989 17:40 | 5 |
| re .15
I too can't seem to find NET: and would like someone to please upload.
Thanks in advance....Steve
|
2837.21 | | BAGELS::BRANNON | Dave Brannon | Fri Aug 18 1989 18:06 | 12 |
| re:.19
Know of any imbedded SCSI drives with ports for 8 Logical Units?
Or any ports on the drive for accessing Logical Units?
Seems like a nice spec, but the implementation of imbedded SCSI
really puts a crimp on the ability to use SCSI as a network bus.
I've heard of 2 and 4 LUN SCSI-to-ST506 controllers, does anybody
sell a 8 LUN controller?
-Dave
|
2837.22 | LUN's | WJG::GUINEAU | Opening the doors of Perception | Fri Aug 18 1989 18:48 | 7 |
| No, no embedded scsi disks use logical units (why would they? Just extra
work for nothing)
I think when things like media changers, printer servers, rs232 servers etc get
popular, LUN's will pick up.
John
|
2837.23 | | BAGELS::BRANNON | Dave Brannon | Fri Aug 18 1989 22:11 | 12 |
| re:.22 (why would they? Just extra work for nothing)
You hit it exactly. Why bother providing expansion capabilities.
So much for SCSI, maybe SCSI 2 will address how to go beyond
8 scsi devices on the bus. LUNs seem like a lost cause.
Unless... somebody makes a SCSI device that can have embedded
SCSI devices as LUNs off of it. I've only seen Adaptec and OMTI
units, the LUNs they talk to are ESDI or ST506.
-Dave
|
2837.24 | | POBOX::ANDREWS | I'm the NRA | Sun Aug 20 1989 04:28 | 2 |
|
I'll try to copy the NET: Stuff to TAPE::, look for it there!
|
2837.25 | more busses | WJG::GUINEAU | Opening the doors of Perception | Sun Aug 20 1989 10:34 | 37 |
|
re more SCSI ID's:
Nope. SCSI II has no extra device support. They provide for a 32 bit
data path through the use of a second cable, but no arbitration
can happen on this cable. So we're still limited to the 8 primary
SCSI devices. IMHO a severe architectural limitation on a growing bus.
SCSI II does claim to be doing away with the LUN field in the CDB (which
allows only 8 LUN's per ID for a total of 64 devices) and recommends
use of the IDENTIFY message. I think this provides 1024 LUNs/ID but
there still LUN's.
You could make a SCSI server thingie which hung SCSI devices off it as
LUN's. The problem is that since the devices behind your server hardware
are true SCSI devices, they are limited by the stub cable length of 4 inches-
makes for a tight box! Unless the server hardware did something funny, this
really isn't feasable.
Now what could be done is for the server to have it's own private SCSI bus
hanging behind it. Now you select the server from the main bus, and specify
an LUN. Theserver goes onto it's private bus, selects the device
corresponding to the LUN you specified and passes the command to it. It then
provides data pass through between the busses. You've just gotten 8*2-1 devices
total (-1 sincethe server takes 1 ID on BOTH busses). MAkes for a bit slower
initil transaction time (arbitration to command phases) but gives you more
inherent true SCSI device capability.
Or you could do what DEC does: Put 2 busses on the system (A bus and B bus,
sorry, no magic bus:-). Here the system figures what device is on what bus.
I'm not sure if Amiga would handle 2 SCSI cards (i.e. 2 A2090's) in one
box (although with the amiga device philosophy, it should...).
Maybe 1 A2090 and 1 A2090(a). You'ld boot off one bus and have a second
just full of devices!
John
|
2837.26 | Just plug 'em in... ;-) | FRAMBO::BALZER | Christian Balzer DTN:785-1029 | Mon Aug 21 1989 04:46 | 12 |
| Re: .25
Yeah, it's possible to plug in more than one SCSI controller into
the A2000. My current record is one GVP, Supra and A2090 at once,
during some tests. There are definite problems with old Cltd
controllers (they don't terminate the bus properly).
Still, not the cheapest way to build a network. ;-)
Regards,
<CB>
|
2837.27 | | BAGELS::BRANNON | Dave Brannon | Mon Aug 21 1989 13:13 | 15 |
| re:.25
thanks for the info, I was wondering what future SCSI had. I've
been seeing ads pushing ESDI as the growth path from ST506. The
main feature being that it was faster, haven't seen much of a push
for SCSI in the ibmpc market (other than pcs sold by DEC).
I read that Transactor article on the NET: device, it sounds like
it should be a lot of fun to try.
At this point, MIDI hardware looks like the cheapest way to hook up more
than two Amigas, plus you could actually use it for MIDI when
you get tired of using it as a computer-computer network.
-Dave
|
2837.28 | | CANAM::SULLIVAN | Steven E. Sullivan | Mon Aug 21 1989 14:27 | 24 |
| RE:.17
> That sharing of the harddisk is possible now with Cltd's SCSI-net.
Ah Ha! Bagels breath! SCSInet is only a rumor from the Cardco charlatians
of C-Ltd!
My AmigaCluster (to which Jeff earlier refered) don't use no "stinkin
SCSInet software." It just uses plain old standard SCSI arbitration for a
standard multi-host environment. I can share disks in a read-only manner
and that is what I do for my system disk. No problem. Manual care and
coordination allow sharing of other disks. No magic and SCSInet (from
C-Ltd) is only very very thin vapor.
There is no controller that I know of that supports being a target as well
as a host. This really messes up a network implementation. The most likely is
the Hardframe and it does not. The hardware is there, but the driver does not
handle it at all.
Using an AmigaCluster is a real feature since it allows me to only manage one
system, games, development disk for both systems. This also save about 18meg
over duplicating the same for both Amigas.
-SES
|
2837.29 | Appletalk? | TLE::RMEYERS | Randy Meyers | Mon Aug 21 1989 16:51 | 19 |
| I was at Systems Eyes Sunday and I noticed that they had for sale
CMI's "multiport" board.
The multiport board has a serial port, a parallel port, and an Appletalk
port on it. The box claims that you can buy the Appletalk network
software from CMI. Since there was no box of Appletalk software on
display next to the hardware, I wonder if the software is still vapor.
Appletalk is the Macintosh network from Apple. It uses fairly inexpensive
cables and high speed serial ports to link everything together.
I remember that CMI announced its Appletalk board back in March at
AmiExpo. I was a little bit amused because there was a hardware
panel (of two or three disk drive manufactures) that I attended
where someone asked "How about Amiga networks?" The drive manufatures
said "Naw, no one is working on it, the Amiga market is still too
small." Meanwhile, down in the exhibit hall, I counted three companies
that were promising cheap Amiga networks RSN. (CMI seemed furthest
along at that time.)
|
2837.30 | What, no RMS :-) | WJG::GUINEAU | Opening the doors of Perception | Mon Aug 21 1989 17:35 | 11 |
| > Using an AmigaCluster is a real feature since it allows me to only manage one
> system, games, development disk for both systems. This also save about 18meg
> over duplicating the same for both Amigas.
But that doesn't give you any kind of file locking if for example, Amiga A
was writing or updating a file and Abiga B tried to read it.
Of am I missing something?
John
|
2837.31 | | CANAM::SULLIVAN | Steven E. Sullivan | Mon Aug 21 1989 20:34 | 9 |
| RE:.30
> But that doesn't give you any kind of file locking if for example, Amiga A
> was writing or updating a file and Abiga B tried to read it.
That was why I said I can share READ-ONLY disks. Locking and coordination
of read/write disks is *user* provided. This is what comes "for free."
-SES
|
2837.32 | oops! | WJG::GUINEAU | Opening the doors of Perception | Tue Aug 22 1989 08:31 | 3 |
| Sorry! Missed that part.
John
|
2837.33 | | ELWOOD::PETERS | | Tue Aug 22 1989 11:40 | 17 |
|
re .29
I have a CMI multiport board and seems to work good. There is
an optional SCSI port ( $50 ) that is a simple ( slow ) port. The
Appletalk software is still vapor. It is expected to ship in mid
September.
The board ships with drivers for the RS-232 ( DB-9 ), the
RS-422 ( DIN - Appletalk ), and parallel port. The parallel port
is output only. They also provide software to redirect standard
I/O to any of their ports.
It would be very easy to modify DNET to any of the serial ports.
Steve Peters
|
2837.34 | Renew the topic | COMET::BELLMJ | | Sat Dec 21 1991 23:00 | 19 |
| Old, old old note that I started (hard to believe). Well, I've been
using ParNet for a while (even though my NetStat doesn't work
properly...could be 2.04) and I like how it works. I use it
sporadically enough to not know much about the details, but it's been
pretty valuable.
However, it's pretty slow. I estimate it being about 3x the speed of a
floppy disk, when using two 7.16 Mhz machines. I'm now used to the
2.58 Mbit/sec transfer rate of the HD+ (EXCEPT NOW WHEN IT'S BUST!) and
ParNet does have some limitations I'd like to find a way around.
Anyway, I'm lookin' at the back of my HD+ and the back of the 3000/16
that'll be mine as soon as I start making money (ha!) and they both
have SCSI ports. And I know the GVP controller supports LUN's. SO....
Anyone hear anything more about a SCSI network driver?
Mike
|