Conference hydra::amiga_v1

    In 1974-1984 there was a "get your MBA" movement.  This 'degree'
    requires that you sell your soul, include in your tool kit a poltice
    and a chicken (ususally a techie).  Your daily smoke screen is to spray
    statistics (poorly gathered and usually wrong) like fountains of truth. 
    This is done inorder to assure no one will ever challenge you on things
    like technical usefullness, feasibility, quality and the like.  Instead
    of competing, making a better product, etc., companies began to nickel
    and dime their customers (with no regard for the customer while they
    were doing skin flinting].  I was working for a company called Computer
    Automation about this time frame and a marketing decree was made that
    we would remove the unit select switch from the floppies because the
    switch cost $1.00(US).  The marketing folks said things like, "no one
    EVER uses that switch!"  Unit select switches aren't there for hourly
    use.  They are there to save your butt when you have to have a
    different numbered unit.  The solution was to replace it with a switch
    which cost about $.50 and was mounted INSIDE the chasis.  You had to
    take the machine apart to get to it.  Eventually it was replaced with a
    foil type switch which when coupled with an LED showed the current unit
    number.  Marketing research had shown that customers NEVER used this
    feature.  However the customers (who probably had NEVER been consulted)
    HATED the new box!
    
    This attitude of nickel and diming was shown to have saved the company
    hundreds of thousands of dollars because they didn't have to stock the
    $1.00 switch, they could instead stock the $.50 switch.  They applied
    it to other areas besides the floppies.
    
    This attitude showed up in cars, computers, just about everything.  And
    customers began to buy Japanese made products which added features and
    gadgets, instead of taking the features away.
    
    Another example of non-technical people making technical decisions (or
    even marketing decisions) without knowing ANYTHING about what they are
    supposed to be making decisions about!
    
    Clark

Um - maybe you should stop buying hardware on a lowest bid basis?

Those are the only times I've run into that kind of problem. Other times,
I get things done the way DEC and Apple do SCSI - each box has two
Centronics plugs, and you cable them together and hang an external resistor
pack off the last box. I even made sure to build my AT mini-tower disk
expansion box to work that way. At this point, I wouldn't buy anything
that doesn't confirm to that.

Net result - I only have to play with resister packs when I fiddle with
the internal configuration of the Amiga.

Similar comments apply to the SCSI ID switches, but I'm not as picky about
that. After all, properly designed SCSI devices don't care, so you should
only have to set it on the initial install.

	<mike

    re: .2
    
    Um - I encountered these problems with a GVP hard-drive
    controller/drive. They bill themselves as the largest seller of
    add-ons for the AMIGA market; not the cheapest.
    
    Um - you seem to have known already what to do. Maybe the average
    consumer is not buying a controller/disk drive to exercise his
    knowledge of the SCSI interface, but because he/she would like
    more space, more speed. I seriously question whether even adding a
    resistor pack externally at the end of the chain makes sense. That's
    like requiring people to figure out where the end of their home-wiring
    is and making sure they put a terminator plug in it. Sure. The only
    reason they don't do that is because of the large number of lynchings
    that would occur when people started frying because their homes weren't
    wired according to some silly spec.
    
    Ken

Interesting note, considering I was just mucking with these recently.
I've got an undergrad hardware background, so I understand what these
things do and why they're needed, but I must admit it certainly isn't
user friendly.  Especially, when it's not always obvious what you're
supposed to do.

I've got a tk50 on my A3000 and I suspected the bus wasn't terminated
properly.  Someone in this notesfile suggested the problem was to remove
the terminator packs on the hard drive.  When I did this, the system
wouldn't recognize the drive was even there.  With or without the
tk50 attached, with or without the tk50 terminated.  Finally, I put
the packs back on the drive and took the pack off the tk50 and now
everything is FINE.  I mean no more intermittent hangs at all. I've
been doing backups every night to reproduce the problem and it
just doesn't happen.  I thought I did the right thing by terminating
the tk50 and that seems to have been the problem all along.  Go figure!

Seems to me there could be a bus approach analagous to the auto-config
spec CBM has defined which could avoid setting unit numbers.  SCSI
doesn't do this though as far as I know.  The termination problem is a
law of nature.  You could design one which didn't require them, but to be
reliable you'd have to slow it down immensely and no one would buy it.
They could put switches on the box, but that doesn't help much really.

-steve

    No doubt some of the storage systems people will weigh in soon, but I
    can't resist getting sucked into this rathole ( ;-).
    
    You can leave the resistor packs off of both drives and cable ends for
    cables up to, say, a foot long.  Anything longer than that and you're
    playing with fire.  The interface will be flaky or not work at all due
    to high speed signals being reflected back from the far end.  This is
    due to the laws of physics, not American attitudes.  Ever had to
    recover a hard disk partition that's been munged by flaky hardware?
    
    All the alternatives you are likely to think of have disadvantages
    which the standards group thought outweighed the inconvenience of
    fiddling with resistor packs.  Mike Meyer's suggestion makes a lot of
    sense in this context.
    
    As for address select jumpers, they're a minor inconvenience, nothing
    more.  Automatic schemes based on cabling don't work well, IMO.
    
    What's needed, actually, is better documentation from host adapter
    manufacturers like GVP and Supra.  I've owned boards from both, and
    find that there isn't quite enough in the manuals to talk a novice
    through step by step.  The manuals are also useless if you try to do
    anything out of the ordinary.
    
    On the other hand, Amiga owners have it easy compared to PC folk.  If
    you want to impress a PC owner, just tell him that your peripherals
    have no address or interrupt select jumpers at all.
    
    Now, given all that, seems to me that there's nothing in the base note
    that couldn't be solved in a five minute phone conversation with a
    notesfile regular, or even with an experienced personal computer owner
    (not necessarily Amiga) over the cubicle wall.
    
    Unfortunately, customizing your computer hardware is not completely
    painless.  If you want plug 'n play, most times you'll have to take
    what the dealer or manufacturer offers, or pay the dealer to set up the
    hardware for you.  This reply may sound hard-nosed, but the disk drive
    industry has, IMO, distributed cost vs. convenience the right way.
    
    Wes

Is GVP selling badly-designed external boxes? In that case, it's to bad
that they're "the largest seller of add-ons" for the Amiga.

As for people being fried because their homes aren't wired to some spec,
that's a cause-and-effect relationship. If the power in your house isn't
wired to spec with the local building code, then you've got no beef with
the code if you get fried. The people you want to lynch in that case are
the builders who built an out-of-spec building. It's when the building
meets spec and you get fried that you want to lynch the people who wrote
the spec.

If you're trying to claim that SCSI is to difficult for a non-technical
person to set up, I'll agree with you. Non-technical people shouldn't
be trying to set up non-trivial SCSI nets; they should get someone who
knows what they're doing to hook it up, or to write out instructions they
feel are clear enough to follow. This is the route people take with VCRs
(which are noticably harder to get set up than a SCSI net). Why do you
expect something that's even more expensive and complex to be simpler to
set up?

The author of .1 seems to have hit part of the problem on the head -
companies are taking cost-cutting measures to get "cheap" hardware out
the door. I.e. - a SCSI adapter/disk package that's plug-n-play. In doing
so, they make a larger percentage of their customers happier - until
those customers try and change thet setup. Said customers didn't know what
they were doing when they bought the hardware, and have no one but
themselves to blame for doing it wrong. [Yes, I blaim myself for getting
saddled with Supra hardware that had internal resister packs and unit
jumpers; I've since sold it and bought better hardware.]

As for the TK50Z, it seems that early DEC hardware isn't quite "true" to
SCSI; the TK50Z and the rz23 being high on that list. More recent
things seem to have changed (anyone tried the TK70 or TK30 on an Amiga?)
DEC hardware also isn't very forgiving of other things not being true to
SPEC. The people to blame in this case aren't the designers of the SPEC;
it's the hardware vendors (i.e. DEC for the TK50Z, and other manufacturers
where they don't follow spec).

	<mike

    re .0:
    your flame.  
    I think you've hit upon one of the more basic ideas that is 
    independant of time or culture.  
    
    There will always be people and businesses like that.  
    You just have to decide if its worth dealing with them.
    
    regards,
    dennis
    

    re .0:
    SCSI termination.  The C= 2091 seems to be much more forgiving about
    termination problems than the Hardframe, just from personal experience.
    
    dennis
    

    
    	First, SCSI was designed before users configured devices. At the
    time, trained field service people were the only ones to setup deviecs.
    
    	As for termination, Physics plays a big part here. The bus needs
    termination at both ends ( the controller and last device ).I like the way
    DEC solved the problem with external bus termination plugs. The idea
    is that controllers/system boxes have a BIG ( ANSI standard ) connector
    for SCSI. The cables go into each device and out the other side. The
    last box has a termination plug on the output.  This seems simple to
    me and the electrons like it.
    	As for SCSI IDs, auto configuration would be nice. But setting
    IDs should be easy. DEC requires all SCSI IDs to be set outside the
    box. There are many way to do this. I like the DEC RAxx and RFxx unit
    number plugs. One of our new SCSI devices will have a thumb wheel
    switch on the outside.
    
    	Most of the problems seem to come from installing "Bricks" ( bare
    devices for internal mounting ). Many of the problems can be fixed
    by removing ALL device termination and going to a termination plug
    at the end coupled with remote ID selection. This is what DEC did.
    
    Unfortunatly once a standard is set you then have to start patching
    ( ask anyone with an IBM clone ). The connectors, ID pins, and
    termination are all standard.
    
    		Steve P.
    
    P.S. If you want to draft a SCSI specification change I can get it
         submitted to the ANSI committe.
    

    re: .9
    
    Is the current SCSI specification available on the net?  I'd like to
    read it.
        John Sauter

RZSCSI::RZ$SPECS:S2R10C00.TXT through S2R10C17.TXT and S2R10CA.TXT (appendix)

It's also on STAR::, I'll mail you a pointer.

john

 re:.0   

  You could just buy a pre-packaged unit or have a dealer install it.
  If you don't like SCSI there are other drive units (although I'm unfamiliar 
  with them) IDE's, MFM adapters, and ESDI which may be available for the Amiga. 
   I believe they all have jumper settings though.
  I think Supra makes an add on unit that has the I.D. switch on the outside.
  I thought GVP did, but maybe not.
  I'm afraid there always will be a learning curve though.
  								Ed

    re: .6
    
    you're quite right about local electrical codes, and who should be
    lynched if a poor unsuspecting home buyer gets fried because his wiring
    doesn't conform to those codes. The problem was I think my location of
    the modifier 'silly'- certainly the codes are not silly relative to
    public safety; my point was that they're silly to the consumer who has
    come to expect that he/she will not have to be concerned with something
    for which they do not have adequate training. Applied to the SCSI
    discussion, what I mean is I as the consumer should not have to (at
    least I have come to think that I should not have to, but that's a
    whole other discussion) worry about things for which they do not have
    adequate training. Or even interest.
    
    As far as SCSI and termination, regardless of the 'laws of nature', and
    I know NOTHING about electricity/electronics, which is just about how
    much I WANT to know, it seems weird to me that this termination can't
    be handled automatically when you plug something into the SCSI port.
    Can't any of you electronics guru's come up with some device that
    automatically pulls off the resistors, or pulls them out of the
    circuit, or SOMETHING, when something is plugged into the port?
    
    As far as apportioning costs versus playback, well that's what
    entrepreneurship is supposed to break free from. One of you could make
    a bundle if you'd pay, with your time, for a design like what I naively
    proposed above. And you hereby have the idea for free - I relinquish
    and all royalties - no joke!
    
    I think frankly we've all been brain-washed by the existing system into
    thinking it's the only way to fly, that every cost-cutting decision is
    justifiable. From within the system, which we all are, this is of
    course true. It's only when you look at it from OUTSIDE that you see
    how much more could be produced/accomplished.
    
    Re: .1
    
    I agree. This note is considerably more perceptive than my original
    flame. Thanks!
    
    Ken

    re: .6  Automatic termination
    
    Apple's Localtalk cabling has a "fat" connector located outside the
    computer housing.  That is where you hook in the cable to the upstream
    station.  If a "downstream" cable is plugged in tothe other hole in
    this connector, then a built-in terminator is automatically switched
    out.  Localtalk terminates a few wires at most.
    
    This mechanical termination switching scheme is unreliable to the point
    of being a major pain.  Imagine that multiplied out to 12 lines or
    more...
    
    Somewhere back in this string the point was made that SCSI drives were
    originally intended for field service types to install, not any random
    person off the street.  Let me offer a homely analogy.  It's relatively
    simple to replace the spark plugs on a lot of cars.  You need a socket
    wrench and maybe a feeler gauge.  There are only a couple of ways to
    mess up:  break a plug, misthread it, or get the distributor wires put
    back in the wrong order.  That's roughly the current state of the art
    in installing SCSI devices.  Now, is it reasonable to expect the car to
    figure out the right order to fire the spark plugs if you get the wires
    put back wrong?  Is it reasonable for you to take all the risk of
    getting the wires back correctly?
    
    As a hardware engineer, I can tell you that in big minicomputers
    probably every dollar spent on components must be justified.  In my
    group (Low End Networks) every nickel must be justified.  In consumer
    electronics, such as TVs and most audio equipment, every penny  must be
    justified.  The disk drive business, for both manufacturers and
    resellers, is cutthroat.  I would not expect ease-of-installation
    features, used maybe once per year, be cost justified without a much
    stronger demand from customers.
    
    Wes

    re: .14
    
    I find it hard to believe an electrical solution to the termination
    problem can't be found. How come you can string floppies together?
    Instead of a mechanical solution, how about an electrical one?
    
    As far as justification of costs, this is precisely the point of .1,
    I think; that costs are justified by people who don't understand the
    product, who don't understand the potential of the product, who don't
    understand consumers, etc.
    
    <flame on again!>
    
    It's just like the argument one constantly hears - you've got to be
    first with the product to make the big bucks. This is used as an excuse
    to justify the weenie-ness of american whatchamacallits - the guys who
    take the big risks (hah!) with their money to finance startups.
    
    This attitude has produced exactly 0% of the innovation represented in
    our products; it is not an attitude that will provide serious
    competition for the Japanese, the Europeans, maybe eventually the
    Russians. THEY believe that what sells is better products. And they're
    willing to spend the money to develop them.
    
    <flame off>
    
    Ken

re .13

As has been pointed out a number of times, the real problem is that you're
expecting SCSI to be user-installable, and it just isn't. It's a major
step above everything else I've dealt with in that direction, but it's
still not there. By choosing hardware that does it right, you make things
easier on yourself.

On the cost-cutting measures; remember that the real reason for those is
to improve the bottom line. If it changes what the users sees, you pretty
much have to justify the changes by showing that you'll sell more units
the new way.

For SCSI, there are two types of external boxes - those done as QAD boxes
that are designed to be plugged in and used; no external termination required.
They tend to have internal termination and SCSI IDs, and are great boxes if
you never intend to add another box. If you actually change your SCSI config
with any frequency, you need the second kind of box. These have external dip
switches (which are a pain, but don't require opening the box), and expect
external termination.

After buying the first flavor (with two drives to start with), and then
seeing the second, I swore never to buy the first type again. That sends
the cost-cutters the message that I don't think there actions are justified.
I again suggest you do the same.

As for the jingoistic comments about this being an Americanism, I can't agree.
VCRs (pretty much the a Japanese product) are universally derided as being
far to complicated; sufficiently so that there's a market devices that program
VCRs automatically. Jaguar's cost-cutting/jingoistic measure of buying only
Lucas electronics rendered it immaterial that the control for the electronics
in the car was a nightmare. Cost-cutting measures that lead to inferior products
are part of a free market. The solution is to refuse to buy the inferior
products.

	<mike

	It's been a while since I flamed.
<flame on>

	The community of people who use computers fall into one of the following
groups:

	1. Those who are technically competant and can configure their machines
	   themselves
	2. Those who wouldn't dream of doing anything to their computers
	3. Those who use machines at work, which are under service contracts
	4. Those who want to do their own configurations but don't want to
	   know any thing about it.

	The number of people in group 4, I suspect, is very small. To quote a 
memorable line from a forgettable movie, "There's two kinds of people in this 
world. And you ain't neither of 'em."  The point here is that everybody would
have to pay the added cost for a feature that only a few people would care
about. This is one particular nit. There are others. If features are added for
every little nit that anyone can think of, the cost will be way out of line.
This is called one-plussing a product to death.

	I don't know of any Japanese, or European, or Russian, or any countries'
SCSI drives that handle termination resistor packs automatically. Check out
your average Asian PC and you will find jumpers and configuration switches
all over the place. Their priorities are cost, cost, and cost. And they seem to
do pretty good volumes that way.

<flame off>

    We have it pretty good on the amiga overall.  I spent several hours
    once getting my wife's clone setup after we installed windows.  The
    mouse, modem, and windows just didn't want to coexist.  Talk about
    a painful experience.
    
    -steve

    <yet another flame on - this is fun! - the last paragraph about the
    Amiga, in case you want to cut to the chase>
    
    There was an American businessman in the fifties, I think, who had a
    Message for American business - you're too interested in money, and not
    enuf in product. Nobody was interested, over here. In Japan, on the
    other hand, his words were heard. He at least provided a direction for
    Japanese industry - quality, product.
    
    America still doesn't hear. Whether it's because of accounting-trained
    MBA's who are being vested with product decisions, which they make
    based on money, or whatever. We don't hear. Actually,
    the replies I've read here so far provide an excellent reason why we
    don't hear - because we have a whole suite of airtight reasons why we
    don't NEED to hear:
    
    1. the consumers won't buy products they don't like. Therefore we won't
    SELL'm. Therefore we won't make money. Therefore we'll HAVE to change. 
    Here are a few problems with this:
    
    a. Simply not buying a product provides no information to the
    manufacturer about WHY the product is not being purchased. It does not
    allow a detailed enuf analysis.
    
    b. If people aren't buying your product, you have a whole slew of ways
    to increase sales without increasing quality - namely, advertising.
    Market the sizzle, not the steak. It's probably a lot cheaper, too.
    
    c. By the time the consumer DISCOVERS that the product has a flaw, it's
    too late; they've already spent the money. What info flows back to the
    manufacturer?
    
    2. Many consumer complaints are from only a fringe group of users; we
    can't afford to satisfy every such complaint; we'd be - what was it,
    oneplussed? to death.
    
    a. If you interpret the consumer-driven nature of product development
    to mean, you wait for the consumer to indicate a massive interest in a
    feature, rather than driving product development yourself, from your
    own common sense, and desire to produce good products, you'll never end
    up getting a good overall view of your product and its market. You'll
    never have a reason to rise above the sales numbers. You'll always be
    vulnerable to startups with a better idea; you'll always play follow
    the leader. In other words, you don't make product, you make money.
    Let me put it another way - a company that DOES oneplus to death, ends
    up so far ahead of other companies they can't even see his dust. He
    knows his product so well, by having been thru so many permutations and
    design attempts, he's a mile away from the competition.
    
    Look at the amiga. It started out with a group of people who were only
    interested in a new, good product - startlingly good in fact - , PLUS
    presumably a group with money, who bankrolled them. The first group
    produced the AMIGA; the second group produced the sale to COMMODORE -
    hey, man, we have to make our money back! It's been downhill ever
    since, altho we started out so far ahead, we're STILL ahead. And of
    course it's been downhill because the people who bought the product at
    COMMODORE are exactly like the money-people who financed it originally
    - main interest = $.
    
    <flame off>
    
    Ken

    re: .19
    
    Well, while I agree with your comments about American business, I think
    you are way off base on SCSI.  As mentioned in previous replies, SCSI
    devices were never designed to installed by someone off the street.  
    When SCSI standards were written, no one expect Mr. Joe Average to be
    getting into his computer that deep.  After all, computers were sold by 
    dealers and the dealer should be doing that.  I don't think cost had
    anything to do with it.  Actually, it isn't THAT difficult to add SCSI
    devices anyway.
    
    FWIW, American computer manufacturers HAVE made very user-friendly
    machines.  Look at the Commodore 64.  Commodore expected to sell this
    through mass-merchanisers and expected Mr. Joe Average to install his
    own peripherals, so they designed the serial IEEE bus.  Just plug the
    floppy in and it configured itself.  Of course, this cost money and
    many people griped about the high cost of the 1541 drives but if you
    want this kind of convenience, you have to expect to pay more.
    
    I think the SCSI is a good comprimise.
    
    Ken
    

    
    picky, picky, picky...  Apple showed how to do SCSI devices in a 
    consumer level personal computer.  You can do the same thing on
    the Amiga if you get a SCSI controller with the DB25 connector.
    Or a 50pin to DB25 adapter cable if you only have a 50pin.
    
    After playing with the homebrew configurations, it may seem to
    actually be worth the extra cost doing it that way, just to reduce the
    pain when you change configurations.
    
    But that takes all the fun out of going to Computer Flea Markets 
    for bargain priced hard disks, streaming tape drives, etc. :-)
    
    Dave
    

    re: 21
    
    could you provide more details on this SCSI controller with the DB25
    connector... or a 50pin to....? Do you know who makes these
    controllers? Does it handle both scsi terminations and scsi-ids?
    
    Ken

    re .22
    
    	Many SCSI controllers support the DB25 connector. I know my GVP
    controller has one.  The DB25 connector was started by Apple. In
    Apple tradition, they ignored the SCSI specification and choose their
    own connector.
    	The DB25 connector contains the same active lines as the SCSI 50 pin
    connecters. They just removed the ground lines ( which lowers signal
    intergrity on the bus ). If you keep the total SCSI bus lenght below
    3 meters ( half the SCSI specification ) The DB25 should be O.K.
    
    	The DB25 connector doesn't change IDs or Termination.
    
    		Steve P.
    

    I had a hard time installing my RZ24 on my GVP. I think it had more to
    do with the dec drive than the GVP. Now both drives are internal to the
    amiga so I had to open up my box every time I tried a different 
    configuration. I found that GVP does not terminate their drive. This
    makes it easy to add another external drive to the bus. I just had to
    plug in a external removable to have it work. That did seem odd to me
    because I was expecting to have to remove the terminateing resistor
    packs, only to find there were non. If all devices were external it
    would be nice but that is not realistic. BTW My dealer installs all
    these things at no cost at time of purchase, to assure they work.
    he has very few unhappy customers in dealing with SCSI.
    

    re: .22
    
    To clarify - Apple went with a DB25 SCSI connector on the computer
    and external boxes which let you set the scsi id from the outside.
    To terminate the cable you needed a SCSI terminator that plugged
    into the end of your SCSI bus.  Each external box had two connectors
    so that it could pass on the bus to the next box.
    
    Simple, end user installable, etc.  But expensive.  With the Amiga
    2000/3000 you get drive bays and slots for hardcards, so you don't
    need an external box.  And when you buy a bare drive it usually
    comes with the terminating resistors.  The controller card normally
    comes with a 50pin ribbon cable, so you have all the parts you
    need to hook it up one drive.  The fun starts when you want to
    add more.  The optimal configuration has the controller card at one 
    end of the bus and the terminator at the other end.
    
    The Apple way makes installation and configuration changes easy.  It
    depends on how often you feel you would need to change things vs. the
    extra costs vs. the potential hassle/grief.  Or let the dealer worry
    about the internal drives.
    
    I bought the 25pin stuff with the idea of replacing my 50pin external
    cable with it, then decided making the change to a working system
    just wasn't worth the potential hassle/grief :-)
    
    Dave

    You may all think we (the Amiga crowd) have it bad, not so, yesterday I
    was trying to get a hard drive working in a true blue IBM PS2.  What a
    piece of c***.  They have no provision for external drives (floppy OR
    hard) unless they come from big blue.  Not a single power connection to
    be seen.  Have any of you tried to add a second controller to a PC?
    it's a DOS nightmare.  How about memory upgrades on a PC?  And yet
    people buy these things like they would be buying bread in a soviet
    supermarket.
    
    Jean
    

    well, the latest installment of the saga:
    
    I called the Memory Location, Xetec, and GVP, and found out how to
    install the xtra drive:
    
    as has been mentioned in other replies, you 
    
    a. open the GVP, which will be SCSI ID 0; take out the resistor packs
    on the hard-drive; make sure the jumpers A0-A2 say '0' - i.e., no
    jumpers
    
    b. open the Xetec; make sure it HAS the resistor packs on the
    hard-drive; set it up to have SCSI ID 1 - i.e., put a jumper on A0, no
    other jumpers
    
    c. close everything.
    
    Unfortunately, it still didn't work. I still got the SCSI error
    $49/$46. At this point, after opening and closing these disks 2-3
    times each, I simply decided to go with the GVP and for the moment
    forget about the Xetec - I can bring it in when I'm willing to be
    without my hard drives for a couple of weeks.
    
    Unfortunately (again) I couldn't close, this last time, the GVP.
    Everytime I tried to close it, a little gizmo that looks like a
    tiny black water tower on three stilts got knocked off one of the
    stilts. This was because the little tower was set very close to the
    hard drive itself, and when I mounted the hard drive, it would knock
    the gizmo off.
    
    So I called the Memory Location and they're going to close the damn
    thing for me.
    
    End of story, for now.
    
    I suppose the AMIGA is much more convenient than IBM; but not knowing
    IBM at all, I can't confirm this.
    
    As far as the Apple, why in the world is it so costly to make a
    separate terminator to fit in the last SCSI device on the chain?
    It's just a little piece of plastic and metal. Come on, guys.
    
    Ken

Re: .*

I would like to join the defense of the SCSI spec.  Internal SCSI devices
are not meant to be "user installable."  Period.  That's all.  They are
components meant to be installed by technicians.  However, they did such
a good design, fumble fingered types like me can receive minimal
instructions and do my own installation.

On the other hand, the snazzy external SCSI boxes are meant to be user
installable.  Almost all of them (well, the ones I've read reviews of
in personal computer magazines, like MacUser) have switches on
the box to set the SCSI id number and external resister packs.  They
are far less hard to install than your average VCR or stereo component.

Installing anything inside of a personal computer or any make is a job
for a technician or at least the technically knowledgeable.  (Wes's
comparison to changing the spark plugs in a car is a very good analogy.)
The guts of a pc are delicate: components are static sensitive, jumpers
have delicate pins, and edge connectors on cards just do not have the
strength that you would want if users were in there every day plugging
them and unplugging them.

When I started looking into getting a hard disk for my Amiga (years ago),
I started surveying the market by trying to see what hard disks were
available and what the vital statistics (size, seek speed, data transfer
rates, ...) were for different drives.  I had a hard time.  Back then, the
Mac II had just come out, and so there was zip demand for internal SCSI
disks.  Most of the market was for external SCSI disks in the pretty
boxes with thumb wheels for setting the SCSI id.  The problem was two
fold:  First, I'd rather have one less box on my computer desk, so I'd
prefer an internal disk.  Second, the pretty boxes with their own power
supply and those neat id switches added about $400 to the cost of the
drive.  Convenience and having someone else assembling the components
always has its price.


Re: .19

>    Look at the amiga. It started out with a group of people who were only
>    interested in a new, good product - startlingly good in fact - , PLUS
>    presumably a group with money, who bankrolled them. The first group
>    produced the AMIGA; the second group produced the sale to COMMODORE -
>    hey, man, we have to make our money back!

Well, *RJ*'s story is that the second group (the backers) ran out of money
and the first group (the techies) were about to be out of a job.  The
company was bankrupt, and the backers were about to lose all of their
investment and the techies (who had been working for stock) were about
to not be paid.  Amiga Inc wasn't sold in order to turn a quick profit
for anybody:  It was sold so that lots of people weren't facing finical
ruin (Jay Miner, the designer of the custom chips, had taken out a second
mortgage on his house to meet the payroll), and so that they company
would survive long enough to produce a computer.

Commodore then did support Amiga Inc with enough resources to get the
job done.  The Sun workstations for each software engineer were particularly
close to *RJ*'s heart.

One the other hand, Commodore did cost reduce the 1000's case, and
didn't permit the addition of a battery backed up clock to the 1000
because "it would cost too much."  (Notice that the battery backed up
clock is still an option on the Amiga 500, even though it is built into
the Amiga 2000 and 3000.)


Re: .25

>To clarify - Apple went with a DB25 SCSI connector on the computer
>and external boxes which let you set the scsi id from the outside.

This was a cost reduction on Apple's part.  A couple of years after the
Mac's introduction, Apple wanted to be able to say that Mac's were
"hard disk ready," but didn't want to raise the cost of the machine
to include a hard disk or the power supply to feed it (or a fan
to cool it).  The solution was to put a SCSI host adapter (disk
controller) on the mother board, but force users to buy external
disks.  The DB25 connector was cheap and sturdy compared to the
(perhaps more proper) 50 pin connector.

>The Apple way makes installation and configuration changes easy. ...
>Simple, end user installable, etc.  But expensive.

Expensive, but it keeps the base price of the machine down.  These days,
Apple assumes that most Mac owners will actually buy hard disks.  Accordingly,
current Macs allow (come bundled with) internal hard disks to real price
of ownership down.

    couldn't resist the final episode:
    
    i brought my battered (by my frequent opening/closing) GVP
    controller/hard-drive back to memory location, where i bought it,
    and humbly asked them to put it together.
    
    "no problem" they said. $35 poorer, i took it home, and plugged it in,
    and booted it up.
    
    this time the disk drive wasn't even recognised; i got the
    'insert workbench 1.3' message. Huh? The controller power light
    was on ok, the switch on the top of the unit that selects between
    a bootable hard drive or not was set to 'bootable'.
    
    i decided to risk opening it up one more time: it turns out the
    technician had forgot to put back the connection from the
    afore-mentioned 'bootable' switch. i did so; works fine now.
    
    ken
    
    PS: this is a revised version of a reply i had originally wrote;
    i deleted a final paragraph that i felt had too much flame and
    too little substance.
    

    i deleted the offending paragraph before i saw your reply.
    
    what's whining to you, to others is an opportunity to produce better
    products.
    
    remind me not to buy anything from YOU.
    
    Ken

    The key is to whine productively (not that you aren't).  You have
    to remember to add a few attaboys.  Like for example, how much
    better scsi is than the typical pc non-scsi controller.  Support
    for up to seven devices is a lot better than just two or even one.
    There's nothing wrong with never being satisfied!  Isn't that part
    of six-sigma?
    
    -steve (the proverbial nit-picker)

    re: .32
    
    thanks! you're right, i was being only negative. a condition i
    sometimes fall into when things go wrong; but from which i eventually
    recover.
    
    without being maudlin (i hope), i LOVE my amiga, and its hard drives,
    printers, terminals, mouse, etc. if it weren't so, i probably wouldn't
    give a sh*t, and would never even bother to complain.
    
    ken

Ken, you've got it backwards

My standards are clearly higher than yours (I wouldn't have bought the
hardware you've got). Obviously, I probably shouldn't buy hardware that
you consider acceptable. I, on the other hand, refuse to buy systems if I
know that better is available and worth the cost difference. Buying hardware
from me would probably improve things for you.

If something isn't up to snuff, you basically have four choices: do without,
buy something better, build something better, or live with what you've got.
I've done them all at times.

If you believe you can do something better than SCSI, do it. Don't complain
repeatedly about it not being good enough - prove that it isn't. If you can
do that, and price it so that it's worth the difference, I'll buy it. I
happen to think that SCSI done well enough to be worth buying is acceptable
(which is _not_ what you've got), so I'm not going to bother trying to improve
it.

BTW, you might note that there are boxes that can communicate with each other
by just plugging them into wall power, from anywhere in the house. That's the
level you should be shooting for. After all, what could be more convenient
than that.

	<mike

    re: .34
    
    In general, I agree with your point that, rather than complain about
    something, do it better yourself.
    
    Unfortunately, I have absolutely no hardware skills whatsover, nor
    the time or inclination to acquire them. If you will remember, one
    of my earlier replies made this point, and asked any of you who DID
    have the skills, to consider some kind of a fix.
    
    I.e., the point of complaining, under these circumstances, is to bring to
    the attention of people with the requisite skills, or people who
    know people with the requisite skills, or ... the problem, which
    may be simply an inconvenience to me, but which may nevertheless be
    easily fixed.  Who knows, maybe one of the manufacturers themselves
    will somehow come upon the complaint.
    
    Another positive result of complaining, despite how obnoxious it may
    seem to people like you who have spent the time and energy to get your
    house in order so everything is neat and clean and generally
    hunky-dory, is simply to keep all of us abreast of possible problem
    areas.
    
    In general, I do not accept the notion that complaint is unacceptable;
    we live hopefully in an open society; my complaint was not vicious, or
    spiteful; it WAS definitely angry. By casting this complaint out into
    the waters, it may help either this problem area, or some other problem
    area.
    
    Ken

I don't believe that complaint is unnacceptable. It's when it becomes
repeated (_especially_ after people have pointed out solutions that haven't
been tried, or reasons why it can't be done) that it gets out of hand. That
you deleted the last complaint indicates you might agree with this, and
consider the last complaint to have crossed that line. Jingoism doesn't help.

The problem with repeated complaints is that you eventually get ignored.
I.e. - Marc Barrett on USENet keeps complaining about the Amiga grahpics
being scummy, and wanting to know why there aren't 24-bit grahpics on the
3000, and DIG, etc. He's got a point - those would be great things to have,
but there are good reasons that CBM doesn't offer them yet. He does similar
trips on different items, and has been doing it for over a year now.

He can't ask a legitimate question anymore without getting flamed. Many
people obviously don't bother to read what he's saying if he has a complaint;
they just flame.

	<mike

    re: .36

    jeez, my first note on this subject was 6 days ago! if that's
    an unacceptable amount of complaint to you, maybe you have an
    unusually low threshold!

    as far as jingoism: i assume you mean reverse jingoism, i.e.
    excessive negative comments about one's own country (since I'm
    born-and-bred american, myself). Well, i made specific assertions
    about what i think is wrong, that can be discussed on their own
    merits. jingoism to me means, 'my country, right or wrong', and
    reverse jingoism then must mean, 'never my country - it's always
    wrong'. That was certainly not my point; i object to our loss
    of a competitive advantage we once had. It used to be america
    who led the way with CONVENIENCES for people - cars, washing machines,
    cameras, tv's, radios, etc., etc. See? When you're concerned with
    CONVENIENCE, when you want more than anything else to provide it
    for your customers, because you know they'll reward you by BUYING
    your stuff, you win. If you have your kind of Caveat Emptor (let the
    buyer beware) attitude, you lose. And that's why I remarked that I
    would never buy from you - I meant you as the designer/producer of
    goods. I imagined you as the designer/producer of goods saying the same
    thing; and I would come out the loser. I interpreted you as
    excusing that type of designer/producer. My own feeling is that those
    earlier americans, rather than evaluating each convenience that they
    might add to their product from a cost perspective, LEAPED at the
    chance to RISK adding it. 

    When you say something like "Commodore has good reasons..." I have a
    kind of sinking feeling, as though I'm stuck in a market that never
    sees what I want, but only sees their bottom line. We didn't get to
    the top that way, and we're not going to stay on top that way.

    When all is said and done, however, I still believe we have great
    ideas for products; that is, our people are still churning out these
    ideas; but fewer and fewer of the people with the money to
    bring these products out, are doing so.

    Ken

Gee, it's only been 6 days? That's greater than one complaint a day. I
don't think that's a low threshhold at all.

I find your attitude somewhat contradictory. You complain because your
hardware is inconvenient, but you reward the people that chose to make it
inconvenient. I can't think of a logical reason for doing that.  The
illogic of "I want the government to think for me so I don't have to"
might apply, but I'm not sure. Since the rise of that attitude and
the fall of american competitiveness happened in parallel, I wouldn't
expect you to take it.

I can see how my not having any sympathy for those who buy substandard
hardware might make you think that's what I'd manufacture. Then again,
that I'm not willing to *buy* something like should be a strong hint that
I wouldn't be willing to produce something like that.

As for ignoring things like marketability and just risking money - have
you considered that there may be a reason that companies that do that
aren't around any more? Developing and marketing something that you
believe will pay for itself when others tell you it won't is bravely taking
a risk. Doing the same thing when you don't think it'll pay for itself
is just stupid.

People asking for a 24-bit board from CBM, or a more convenient HD spec,
are asking others to risk money. I don't have any sympathy for those
people. You might as well complain that you can't buy tickets to the
moon. Those who design and build their own versions of these things have
my admiration - they are taking the kind of risk you want them to. Like
ALS and competitors, they're building their own way to what they want.

	<mike

    re: .38
    
    well, i think i understand your attitude now.
    
    first of all, i want to say that i am not content to simply complain;
    my field is software, and at home, on my amiga, i am trying to
    create products that in fact have the kind of convenience, appeal,
    useability, attractiveness, etc., etc. that i myself would like to see.
    i don't know how many times i have to say i don't have hardware skills.
    if not having hardware skills means i can't complain when hardware
    lacks features i'd like, well, that's your opinion, not mine.  if
    a hardware person who had never programmed a line in his life
    complains about software, i listen. and if i have a convenient answer
    for him, about why it HAS to be so, nevertheless i listen AGAIN.
    maybe i was wrong, maybe it in fact IS possible to improve. if
    he's angry enuf, i listen twice as hard.
    
    i also don't know how many times i have to explain that i didn't buy
    the hardware knowing it had the inconveniences i later found; i guess
    you don't hear what you don't want to hear. you in fact have the
    attitude, as i said before, of caveat emptor; if you have so little
    respect for the average consumer, then i continue to believe it would
    be a mistake to purchase a product you contribute towards bringing to
    market. but at least you've let us all know.
    
    i guess your caveat emptor attitude is based on a feeling that
    capitalism is kind of like, survival of the fittest. you probably
    feel that greed, thru an invisible hand in the marketplace, 
    miraculously brings us all good things. quite understandable, since
    that's what we hear everyday. and you feel the government's role,
    rather than protecting the consumer, is to protect those who prey
    on him or her, since after all, that invisible hand helps us all.
    You have the peculiar idea that a marketplace doesn't need to
    be honest, that consumers don't need to be informed, etc.
    I guess you must feel that since you've figured it out, and it
    cost you an unpleasant experience, then that's good enuf for
    everyone else.
    
    Finally, you seem to have the strange conception
    that new products come to market because people like you make
    rational, calm, well-considered evaluations of profit, market
    share, etc. Well, forgive me, but people like you have nothing to
    do with bringing new products to market, at least not the kind
    that are really new, really change things for people. On the
    other hand, rocky 2, rocky 3, rocky 4 - well, those are i think
    well within your capabilities.
    
    i guess you feel your logic, or what you call logic, which you
    have apparently mastered, or think you've mastered, protects YOU
    from getting hurt in the marketplace, so the devil with everyone
    else. 
    
    ken

re last few:

Gee, this used to be a *friendly* conference.

    It's beginning to sound like usenet. Next unseen.

    I always felt that a good dealer would go a long way in avoiding the
    problems that are being discussed in this note. If a consumer has a
    problem with setting up hardware then the dealer should be a source of 
    help. In the case of scsi controllers, it appears that these type of
    products should be installed when purchased. I have seen many cases 
    where this happens. Then if a desire to install another device arises
    and the person is not comfortable doing it himself he should seek help.
    We live in a society where it is the norm to pay for assistance. I had
    to fight with my RZ24 to get it to work, but if I had wanted I could
    have taken it to have it installed. Another solution is to buy
    carefully upfront. I don't look at it as GVP's fault that I can't
    handle adding a second drive internal to my Amiga.If I wanted to all I
    had to do was plug in a Syquest external with no hassle at all. 
    								bill

 One of the basic questions of any non-hardware oriented buyer should
 be, "Can I install this myself?". If the seller asks for your background
 and says yes, then the product should match that requirement. If the
 seller says no, then buyer beware. If the buyer doesn't ask the question
 or doesn't get a *firm* yes from the question, then the buyer should
 make sure that a technically competent person is available, or purchase
 something different.

 With capitalism there will always be at least two solutions to a problem/need.
 If someone builds an easy to use product someone else will build a less
 friendly but cheaper one. If the difficult solution comes first,
 someone else will build the easy to use product later. This is standard
 economics/marketing etc. 

 A note about notes conferences:  It is well documented in psychology
 circles that many shy people, who would never complain face to face,
 are comfortable flaming away "electronically". This is neither all
 good nor all bad, but it helps me understand what may be going on here.
 And some people kick their dogs when they are mad at someone else.

 Keep it lighter and less personal please. 
 Dave

    YA GETS WHAT YA PAYS FOR!
    
    If your dealer is more expensive than mail order, it may be because he
    provides the services YOU are not capable of performing.  I buy bare
    drives and controllers because I can hook them up myself, I can even
    get better prices at the dealer just because he never gets hassles from
    me (hardware wise).
    
    Jean
    
    

Yes, I heard several times that you had already purchased the hardware.
You never said you couldn't return it, or that you even tried. You just
complained that it wasn't good enough. The first round of hard disk I bought
was sufficiently substandard that I returned it.

I don't know how you got the strange ideas about what I think a marketplace
should be like. I do think a marketplace should be honest - but what we're
discussing is a quality issue, not an honesty issue. I also believe that
consumers should be informed - after all, I've been saying all along that
I don't have much sympathy for those making uninformed purchases. I get the
feeling that you don't think the consumer should be informed, and that
everything should be perfect for them no matter what they buy.

I can see where you got the idea that I think new products come to market
because people make rational, well-considered opinions on what to bring
to market. That's not quite the case, though. As you point out, that
leads to a series of minor tweaks on old products. Revolutionary new
products come to market because people who believe in them work to bring
them to market. The don't come to market because people who don't believe
in them are nagged into doing it.

BTW, it's not "the devil with everyone else", it's "the devil with anyone
not willing to work for themselves." If you don't care enough about what
you buy to bother doing anything about making informed decisions, why should
I care about what you buy?

	<mike

    re: .45
    
    The Xetec I originally bought was fine as my one-and-only hard disk. It
    came pre-formatted, bootable right out of the box. I plugged it in
    and it ran like a top. At least for a while.
    
    At this point, since it met my original specs, i kept it. I had
    originally bought it on the advice of amigaworld, which had a
    big review of various hard disks, and said the xetec was much the
    fastest.
    
    to be honest, i do not remember seeing a word about how easy or hard
    it was to change your configuration AFTER you bought it. maybe it
    wasn't there. maybe it was and i missed it.
    
    Finally, the xetec controller died. I decided to get a new controller.
    I decided to get a GVP, with its own hard disk.
    
    This is when I started to run into problems. BTW, I can't return
    the Xetec controller since the warranty is long since expired.
    
    This is the point at which I have started to complain.
    
    Let me try to understand. Are you saying that you don't give a damn
    about my problems with the SCSI interface because over a year ago,
    before I had any idea I would need to change my configuration, I didn't
    have the foresight to look into this, at the time, theoretical issue?
    So now, it's sort of "too late"?
    
    Ken

You take your system with dead controller to the dealer/service center and
say `what will it take to fix it?'.  They might respond with `gee, that's an
old controller, are you sure you want to spend money on that instead of buying
a new controller?'  To that you respond, `what does each option cost?'.
They then quote you a price for each option and you chose one and pay them
only when they have it working.  If they screwed up in estimating the difficulty
of getting the older scsi drive working with the GVP, they eat the cost, not
you.

One of the interesting things about this topic is how it relates to the NMS
(New Management System).

I've recently been working with someone who has been working issues around
HPPI which is a high speed comm channel.  He's working on getting a HPPI
adapter built for the XMI by an OEM.  His business plan is to develop the
expertise so that he can deliver integrated HIPPI support to customers.
None of the HPPI hardware would be manufactured by DEC, and all he would
do is buy the hardware and pass it thru to either the customer or to another
DEC organization.  Obviously, he has to add a fee on top of the OEM price
to pay for the service.  He's taking the attitude, `I don't care whether you
buy through me or not'.  The reason that he feels comfortable taking that
attitude is that he won't talk to anyone about their problems making things
work unless they bought thru him, even if its another group within DEC, and
he recognizes that there is so many variables around HIPPI, only people who
spend a lot of time understanding HIPPI and all the variations can reliably
get it right the first or second time.

Likewise, the reason that a dealer can charge $50-75 to do this kind of upgrade
is that he knows that there are so many variables, that few people will be
lucky enough or have enough knowledge and experience or have the resources
to get it right without an unusual amount of effort.

Replacing your controller is no more difficult than
 - replacing your car stereo with new radio+cd player+speakers
 - replacing your water heater
 - replacing your garbage disposal
 - replacing the brake pads on your car
and lots of people do it themselves.  Are you going to complain about the
water heater manufacturer bad design when it isn't quite the same dimensions
as the one you're replacing and you now have to figure out how to make the
connections?  All of these jobs are really basic and lots of people do them
all the time, but that doesn't mean that 50% of the jobs don't involve resolving
some sort of incompatibility.

    A stupid question :
    
    <flame off>
    
    	Can you have two scsi controller in a Amiga, one controller
        Auto-boot, but not the other ?
    
    <flame on>
    
    							Erik

    re: .48
    
    my xetec docs talk about hooking up 'networks' via the scsi interface.
    i seem to remember this is more than just an ami, a controller, and
    a bunch of hard drives. i'll check it out.
    
    ken

    
    re .48
    
    >> Can you have 2 scsi controllers, 1 autoboot, 1 not ?
    
    Yes. I have run that configuration.  In fact, the autoboot ROM(s) may be
    in either, or both.   I have 2 GVP controllers and GVP suggested that
    pulling the boot ROM out of one of them would speed up the boot.  I
    tried that but couldn't perceive any difference in boot time so I left
    em' both in (at least I'll know where to find it).
    
    Frank
    
    

    re: .48
    
    I'm not sure what you mean by autoboot.  Each of the possible 6 devices
    which connect to a given scsi controller can be either autoboot or not.
    
    Do you mean auto-config?  I can't think of a reason you wouldn't want
    the controller to auto-config.
    
    To have two controllers you need to set them with different LUNs I
    believe.  I'm not sure what the LUN range is but I bet it's at least
    0 and 1, and probably 0-7 :-).
    
    -steve

    hey, i had the docs in my briefcase! i quote:
    
    "setting up a network of SCSI drives and computers is in theory no
    different than a simple one-computer-one-drive system. The same SCSI
    philosophies must be applied: all devices must be arranged in a chain,
    and the two end devices must be the only ones with terminators
    installed. Here's a typical setup:
    
    <simple picture of two hard drives, two computers>
    
    ken

    
    Re .-1
    Steve,
    
    I guess I don't fully agree with the following -
    
    > To have two controllers you need to set them with different LUNs I
    > believe.  I'm not sure what the LUN range is but I bet it's at least
    > 0 and 1, and probably 0-7 :-).
    
    Both of my contollers are the same LUN.  Since each controller 
    is a separate device (by option slot) as far as the Amiga is concerned,
    autocongig works fine.  Since each controller is controlling it's own
    SCSI buss, there is no problem.   
    
    I have no idea if it's rational to have 2 controllers to the same 
    SCSI bus.
    
    Frank
    
    
    
    

    
    re .48-.53
    
    	It seems we are mixing terms here.
    
    	SCSI controller  - PC board in system that support multiple drives.
    
    	SCSI device	 - Disk, Tape, ... units on the SCSI bus
    
    	SCSI ID          - Unit number of SCSI devices ( range 0-7 )
    			   SCSI controller normally uses ID 7
    
    	SCSI LUN         - logical unit number, old SCSI feature that
    			   allows for a multi unit device, This feature
    			   is not used today.
    
    re .48
    
    	Yes, I have run GVP and CBM controllers together in an A2000. I
    	have also run a GVP controller in an A3000. I have run a GVP series
    	I and a Series II controller in the same Amiga.
    
    		Steve P.
    

    re .48
    
    Yes, you can have multiple scsi controllers in a 2000/3000.  I have
    2 ICD's and a GVP.  The ICD's will boot the last "bootable" drive
    found, so you can adjust your configuration by setting the type
    of drive in the RDB (BOOT, MOUNT, SKIP). The GVP is the last controller
    on the bus, so when I enable auto-boot on this controller, it takes
    over and boots my "emergency drive" (i.e. i screwed my regular boot
    disk).  Rule of thumb, last controller on the bus that has a bootable
    drive wins the boot.
    
    This situation is different from that posed in the last reply, which I
    assume implies networking two Amigas using a single scsi bus.  This
    requires that the host adaptors be set to different id's (7 and
    probably 6).  The only advertised Mfg of a scsiNet that I know of is
    the IVS TrumpCard Professional.  I don't have any experience with this
    controller.
    
    As discussed in previous replies, once you add more than one controller
    you have multiplied the complexity of your configuration. If you're
    not comfortable with configuring a single controller, multiple
    controllers may give you quite a headache...
    
    
    John
    
    
    
    
    

    re: -.1
    
    I would like to try that.  I have two GVP controllers and didn't think
    of using both, one for a tape and one for disk(s).  I thought since
    GVP's controllers answer at SCSI ID 7 there would be a conflict.  I
    DIDN'T take time to think through Amiga controller addressing.  Thanks
    for the idea, Frank.
    
    mark

    Sorry for the misinformation, I guess my memory was failing me.
    I thought the LUN business was how a friend of mine had this set up,
    but that was a couple years and quite a few brain cells ago.
    
    -steve

RE:56,57

The scsi unit address of controllers on the same scsi bus must be different.
This is a SCSI issue and has nothing to do with Amiga device addressing and
autoconfig. 

Also, if you have two controllers in the same amiga on the same scsi bus, better
be very very careful about accessing the same partions from different 
controllers. You can trash drives very easily because the amiga sees the
partitions through each controller and never gets the idea they are the same
device. Oops! I do not recommend more than one controller per scsi bus, and
especially in the same machine. It is a pretty lame excuse for a real network
like Ethernet, though it is quite inexpensive.

Another reason scsi on the amiga is pretty lame for a network is that no
controller (that I have been able to locate) fully supports target mode: e.g.
it can not be the receipient of an unsolicited transfer. This is different from
being re-selected by a drive that had to do a seek.

	-SES

What!  No AmigaClusters??

I love this statement!!!
    "setting up a network of SCSI drives and computers is in theory no
                                                          =========

re:58

>Another reason scsi on the amiga is pretty lame for a network is that no
>controller (that I have been able to locate) fully supports target mode: e.g.
>it can not be the receipient of an unsolicited transfer.

There's more to it than that.  I don't know enough about scsi to know what
it specs, but if you looks at SCS, DSSI is a good implementation, not only
do you have `controllers that support target mode', you also have datagram
and message service.  Message service isn't really required, but you must
at least have datagram service between controller so that it can be useful.
If you had that, then host systems could use the datagram service to exchange
information about what cooperative operations are to be done and where in
memory DMAs should be done so that one of the two nodes could actually
initiate the data move between hosts.  One of the features being added
to VMS is IPC using named buffers and this has the potential for processes
on multiple nodes exchanging data at the rate of 5-10 megabytes per second
on CI and 2-4 megabytes on DSSI with very little load on the CPU.

Of all the operating systems around, either mainframe or PC, I'd say that
the Amiga could very well be the second best choice for clustering.  I haven't
yet had enough time to look into how AmigaDOS supports file sharing, but it
can't be any worse than unix, but AmigaDOS is much better designed than unix
in terms of overall architecture.  (Does OSF/1 support dynamic library
activation?)  And from the standpoint of being able to do the development
work, loadable drivers and stuff are really important.  Wow!  Another Amiga
project to add to my list of things I'll never have time for would love to
work on.

How about this flavor.  I have been kicking the idea around.

Adding a SCSI controller on the bridge board to the same SCSI bus as an Amiga
controller.  I would use a separate drive for the Bridge board but hopefully 
would be able to access it through the Amiga via Cross Dos.

Any comments.

Alan

    Would this work?
    
    A2000 with hardframe with SCSI drive(s) <----> A1000 with SCSI cont.
    
    The idea would be to get programs and data from the SCSI drives in the
    A2000 without saving anything back.  This way I could load very quickly
    and do the saves to floppy.
    
    Any comments
    
    Jean
    

    re .62:
    I've used A1000 with Phoenix SCSI controller connected to A2000 with
    2091 SCSI controller to move files to/from the A1000 from the A2000
    side.  However, both systems assume they are the only owner of the
    file system so don't try to access the files until after you reboot.
    
    dennis
    

A long time ago I had a Cltd SCSI500 and they mentiond just such a configuration
But to keep one system fromn writing to another systems disk you would mount
the other systems partition or disk read only.  So each Amiga would have their
own Read/write disk and mount the other amiga disk/partition as read only.  

If you do not do this you have to do a lot of diskchange to keep thedisk update 
current.

CLtd - that was it.

The Sysop of a local BBS uses those. He had a CLtd controller on his A1000
that ran the BBS, and a second one in his 3000 so he could read/verify
uploads without going to the BBS. He may even have had the upload and
main areas in different partitions so he could move stuff "through" the
3000, and only have to fiddle with the BBS files on the BBS when the new
upload area got full.

	<mike

I tried the C-Ltd approach outlined in the last 2 replies. It can work to an
extent. C-Ltd controllers never supported reselect, so all read/write operations
were atomic at the drive level, simplifying the interaction between systems
and slowing down overall operation. The C-Ltd controller was kind of slow 
anyway so this was not a problem.

What was done there could be done between any scsi controllers that have 
configurable addresses. For instance, I used a HardFrame and a C-Ltd to 
implement this configuration. 

The comments about diskchange and related problems are very real. This was a
MANUAL file system cooperation/interlock. By using partitions as read only
by both systems maximum safety could be obtained and operation was stable.
With one system able to write a partition there was risk the other system(s)
could use a bad cached directory block or pointer and get trash resulting
in a crash, or worse, undetected wrong data.

With both systems able to write a partition it was very dangerous, for obvious
reasons.

I tried to make this work for quite a while and decided it was not worth the
effort and headaches that it took.

C-Ltd often refered to some software product called SCSI-NET. I never saw it
and from the information I got from their BBS neither did most anyone else. 
Those who seemed to think they did seemed confused with the multi-controller
setup above.

C-Ltd was not surpassed in their ability to support hacked scsi implementations
with adaptec and omti contollers. The adaptecs were less failure prone and the 
Omti's were faster and more featureful. Both adapted st506/st512 disks to
scsi.

	-SES

We need a distributed lock manager for the Amiga.  I had a customer way
back in VMS 3.x and dual ported RM05's with the same problem.  wish list 
item for amiga dos 3.0?

    So...what is the differences between the SCSI net discriptions and the
    IVS SCSI-NET implementation?
    
    Just curious.
    
    Clark