Hard Drive alternatives for DEC/Alpha Systems.

[dB LEE]

Hello All,
What's the consensus in the comnmunity about
using non-DEC drives on Alpha/VMS systems?

I've got 3 AlphaServer 400s running 7.3.
I need to install larger hard drives and
was wondering if generic SCSI drives can
be used instead of DEC parts.

I found that I can initialize, mount and boot
a 2GB Quantum Viking drive that I gleaned
from an old Macintosh. It seems to work fine.

But when I tried this with a 9GB Compaq U2,
jumpered for Narrow/SE and connected through
an SCA to 50pin adaptor, I had problems.
These seem to be SCSI buss problems that
affect mount verification but I've not tried
to investigate further.

Are non-DEC/Factory drives commonly in use or
is this approach a waste time?

-Dave

[Steven Schweda · Accepted solution]

I've been using fairly random (cheap, eBay)
disks in all my VMS systems for years. My
(now disused) AlpSta 200 4/233 systems have
these system disks currently installed:

RZ1CF-CF (Fujitsu MAB3045SC) (How did I find
a genuine DEC disk? It's a miracle.)

IBM 36L8772 (Seagate ST39103LC)

Currently in my XP1000 systems:

Fujitsu MXA3182SC (system)
Fujitsu MAB3091S SUN9.0G (system, data)
Fujitsu MAB3091SC (Tru64 data)
Compaq/Fujitsu MAH3182MC (system)
Seagate ST373453LC (data)
Seagate ST19171WC (Tru64 system)

(And then there are the VAXes...)

Most of those are SCA-80, connected through
SCA-to-50/68-pin adapters, or in external Sun
611 boxes. I can't guarantee that you'll
have no problems with any random disk, but
I'd also be a bit amazed if a healthy Seagate
or Fujitsu disk had any real problems.
Similarly, SCA adapters (which may cost
almost as much as the disks I buy) seem to
cause no particular problems.

Having had a bad experience long ago with a
Quantum Fireball -- Who picked that name? --,
I've tended to shun Quantum products, and
some loud Fujitsu bearings have tilted my
preference toward Seagate, but, clearly, I'm
still using both of those.

I tend to look for "new" drives on eBay, like
the small pile of ST318404LC drives which the
UPS guy just delivered -- roughly $15 each.
I'd tend to avoid anything much smaller, as
I'd expect, say, a 2GB drive to be pretty
old (and/or slow) in comparison.

> a 9GB Compaq U2,

Actual model?

> jumpered for Narrow/SE [...]

There are jumpers for that? Not on anything
in _my_ junk collection.

[dB LEE]

Steve,
thanks for the detailed reply!
Now I'm encouraged about this.
I've got a bunch of Compaq (Fujitsu ?)
MAH3091MC 9.1 GB drives (Ultra 2, SCA)
that would meet our needs very nicely if
I can get them installed.
FWIW: They have jumpers to force Narrow & SE.

When initalizing this drive should
I be using any special parameters?
Right now I'm just doing:
$$$ initialize dka0:

[Steven Schweda]

> FWIW: They have jumpers to force Narrow & SE.

You win on that one, but the default
(automatic) settings would probably be happy,
too.

> When initalizing this drive should
> I be using any special parameters?
> Right now I'm just doing:
> $$$ initialize dka0:

The defaults are normally ok, but there are
situations (very large number of files, ...)
where non-default parameters may be better.
The defaults have always worked for me, but
that may not say much.

When buying potential junk, I normally run a
drive through a Solaris "format" and
"analyze" sequence, to see if the drive gains
any defects over a few hours. Some kind of
simple BACKUP /VERIFY exercise would probably
be good to build confidence in the drive's
VMS compatibility.

People who use only supported hardware are
all sissies, I claim.

[Guenther Froehlin]

"50pin adaptor, I had problems."

C'mon...I think you can do better and tell us the symptoms of the problem. And a bit more how you plugged it together.

/Guenther

[Allan Bowman]

I have used numerous different drives on my various MicroVAX systems without any problems. When I have purchased drives on eBay, I have found about a 25% failure rate but the prices make it worthwhile.

You don't mention setting the SCSI ID on the drive. If there is a conflict, you will see SCSI bus problems. You need to make sure the drive is set for an ID not in use - and don't forget that the internal controller uses an ID (usually 6 or 7 on the MicroVAX systems - not sure about others).

Allan in Atlanta

[Mark...]

Hi,

As far as I am aware the difference between a "DEC" disk and a manufacturers disk is the firmware.

Digital created there own f/w to optimize the performace of the drives for vax/alpha vms/Tru64 etc. Therefore if you bought a "supported" drive that did not function correctly in a DEC product then engineering would resolve the issue.

If you just put in a manufactures drive + factory f/w and it works then fine, as you have found.

What you probably have not noticed is that when looking at error logs you would get more info from a "DEC" drive than from a standard drive. If you had a service contract and logged a call for a replacement drive then it would not have been replaced as not like-4-like etc. Memory was another area where customers used to put in non "DEC" memory [kingston, dataram, etc] it all worked OK but if you logged a fault call and it was memory related then the non DEC memory did not have the additional features to help track down memory related problems so you would be advised to replace with DEC memory. The non DEC memory would typically not be covered on the contract even if the rest of the system was.

DEC were not unigue in this - all manufactures [even today] want you to stick to branded parts be it computers, printers, cars etc...

You pays ya money...

Mark...

[Steven Schweda]

> You pays ya money...

And you listens to the stories, and you
believes the ones you likes.

> [...] the non DEC memory did not have the
> additional features to help track down
> memory related problems [...]

Sounds like nonsense to me. DEC may not have
been able to translate an ECC syndrome value
into a chip location if they didn't make the
board, but I once spent a few minutes on the
phone with a helpful fellow from National
Semiconductor, who told me which (64Kbit?)
chip needed replacing on a (1MB?) board which
was getting ECC errors logged on a
VAX-11/750. (Conveniently, there were a
couple of spare chips on the board, in
sockets labeled "SPARE", so it was an easy
repair to make.) Compatible ECC memory _is_
compatible, even if it's not identical.
Sales would be pretty slow otherwise.

It's understandable that DEC wouldn't support
(some) non-DEC hardware, and removing it aids
in blame assignment (which ia often
important), but lets not get carried away
when contributing to the oral tradition.

[Mark...]

Hi

ref previous [above] note

ooppss sorry, there was me forgetting that everyone is blessed with your obvious technical skills - my, WHY do we need forums when everybody knows the answers ??? and can fix it themselves ???

Next time anything of mine breaks I'll just call the relevant manufacturer dirrectly, summon the tech department, and I'm sure I'll be able to fix it....

I think you are overlooking the point - the whole idea of forums is to assist and help people who wish to help themselves. I was just trying to give some background info as to why disks/mem sometimes did not function and to the reason why.

If you don't wish to accept another point of view then that is fine by me. If you think my info is invaluble and increasing "myths and legends" then that is fine by me as well.

Ya trys to help people and they can pick and choose what they wish to believe...

Mark...

[Mark...]

Steven

Sorry again, I forgot to ask.

Being as I am obviously misguided and I do not wish to extend the â lets not get carried away
when contributing to the oral traditionâ theme, I [for one] would be most interested to know why manufacturers like DEC produced there own f/w for disk drives when clearly it was not required then or now ???

Iâ m sure you will be able to put me straight so that I do not continue to propel any urban myths to greater levels.

Mark...

[Steven Schweda]

> WHY do we need forums when everybody knows
> the answers [...]

So that we can have friendly arguments
between the people who believe that they have
the answers and the people who believe that
they have different answers, apparently.

> Next time anything of mine breaks I'll just
> call the relevant manufacturer dirrectly,
> summon the tech department, and I'm sure
> I'll be able to fix it....

And, if you paid for your broken gizmo what
we paid for that Nat Semi memory board, and
had its owner's manual with the tech support
phone number, then you could probably talk to
the expert, too, even a few years after they
stopped selling the things. And if the
repair involved no more than swapping a
couple of 16-pin DIP DRAM chips, in sockets,
which required only a small screwdriver and a
bit of careful prying (and the knowledge of
which chip had gone bad), then you (yes, even
you) could probably handle the repair itself,
too.

> [...] that is fine by me.
> [...] that is fine by me as well.

Glad you're happy.

> Ya trys to help people [...]

With non-facts? How does that help?

> [..] why manufacturers like DEC produced
> there own f/w for disk drives [...]

Knowing nothing, I assume that there are
situations, perhaps involving multiple host
access, where the stock firmware misbehaved.
As I said, "I can't guarantee", while DEC,
Compaq, and HP do exactly that. I also know
how many firmware-related disk problems
I've had in many years of running various
bargain disks on my various VMS systems (that
is, none).

Perhaps I've forgotten. Who dragged a bogus
story about non-DEC memory into a disk
discussion?

[Mark...]

ouch!

yes, I do have a sense of humour and I like a debate, however, how about answering my question about why manufacturers develop there own f/w which you seem to insist there is no reason for ??? As I said in my last note - I don't wish to propel an urban myths.

I only used the memory as another example. I never said it wouldn't work, as I know fully that it does.

As for fixing things, well I used to go down to COMPONENT repair myself WITHOUT any bodies help. I don't need to enter any competitions on that, thanks.

Mark...

[dB LEE]

Gentlemen, gentlemen!!!
Thank you for your vibrant discourse. :-)
In the context of my query you have,
generously answered my question:
Non-DEC SCSI drives are a viable and widely
deployed alternative for the DEC Alpha platform.

Now I have an interesting bit of discovery I hope you will find useful, especially you, Guenther, who expressed curiosity in my SCA to 50 pin adapter...

"A tale of 3 adapters"

I have 3 different SCA to cable adapters:
XPL-065F SCA to 68 pin cable.
MC-80 SCA to 68 & 50 pin cable.
XPL-065-C SCA to 68 & 50 pin cable & and optional "9 lines terminator".

Using the XPL-065F adapter I connected a Compaq/Fujitsu MAH3091MC (9.1GB, U2, SCA) to the wide SCSI cable of an Alpha Station 500 running VMS 7.3. I was able to initialize, mount and perform $ BACKUP/IMAGE/VERIFY operations to and from the drive with no errors reported by the $ SHOW ERRORS command. I was also able to install the minimum VMS system and boot off the drive with no errors or warnings reported from the console.

Next, with the MC-80 adapter, I connected the same drive to the narrow SCSI cable of an AlphaServer 400 running 7.3. I set the Wide-Narrow jumper on the drive to Narrow. I was able to perform the same tests as before, with no errors.

Finally, with the XPL-065-C adapter (with the 9 lines terminator removed), I connected the same drive to the same AlphaServer 400 and found I was not able to initialize the drive:
(note, test drive is dka300, system is booting off dka600)


$$$ initialize dka300: DATA1
%SYSTEM-I-MOUNTVER, SABKUP$DKA600: is offline. Mount verification in progress.
%SYSTEM-I-MOUNTVER, SABKUP$DKA600: has completed mount verification.
%INIT-F-MEDOFL, medium is offline
$$$
$$$ show error
Device Error Count
PKA0: 7
SABKUP$DKA300: 1
$$$ initialize dka300: DATA1
%SYSTEM-I-MOUNTVER, SABKUP$DKA600: is offline. Mount verification in progress.
%SYSTEM-I-MOUNTVER, SABKUP$DKA600: has completed mount verification.
%INIT-F-MEDOFL, medium is offline
$$$ show error
Device Error Count
PKA0: 9
SABKUP$DKA300: 2
$$$

These failures, apparently caused by this XPL-065-C adapter, were what initially prompted me to post this question on the forum. Examining the adapter I notice it is wired somewhat differently from the others: all the SE ground pins appear to be tied to a common ground trace that runs between each connector rather than handled individually. Certainly not compatible with LVD signals, but maybe OK for SE? I used these adapters with an Adaptec/PC architecture and they worked fine.
Perhaps there is something in the DEC SCSI interface that is different from the Adaptec/PC architecture... I wonder what?

Hope you found this long post interesting, I certainly have.
Thanks for your help.
-Dave

[Steven Schweda]

> [...] how about answering my question [...]

I thought that I had. Look for the paragraph
after:

> [..] why manufacturers like DEC produced
> there own f/w for disk drives [...]


I assume that my (simple) requirements are
easier to satisfy than those of the full
range of DEC/Compaq/HP systems and
configurations.

> I only used the memory as another example.
> I never said it wouldn't work [...]

As I recall, what you said was:

> [...] the non DEC memory did not have the
> additional features to help track down
> memory related problems [...]

which _still_ sounds like nonsense to me.

And I never said that you said that it didn't
work, but if it _were_ actually missing those
mystery features, than a good argument could
be made that it _didn't_ work (in a
compatible way).

> As for fixing things [...]

Glad to hear it. As I recall, _you_ were the
one who seemed somehow troubled by my ability
to get help from a manufacturer. I try to
get help where I can when I need it. I just
find facts more helpful than fiction in most
cases.

[Mark...]

Try again "Knowing nothing, I assume that there are..." is not an answer.

Please answer the question without going into another essay that means nothing, just smoke & mirrors on your side.

Mark...

[dB LEE]

Non-DEC SCSI disk drives are commonly used
as an alternative to DEC branded parts.
The creator of this thread experienced
difficulities with a certain type of
SCA to 50 Pin adaptor but was otherwise
able to employ non-DEC disk drives in his
systems.

Thank You.

[John Gillings]

re: Mark's question

> [..] why manufacturers like DEC produced
> there own f/w for disk drives [...]

DEC went even further than just firmware for disk drives, they even produced their own version of SCSI, called DSSI, and a whole class of even more heavily modified disks to connect to it - the RF drives. Each RF drive had a tiny cluster aware operating system inside it, and gathered statistics on every I/O operation ever performed. Great technology, but proprietory, and therefore evil in the eyes of the market.

Why go to the expense? Because the standard off-the-shelf components were perceived as being not up to required standards.

Also, back in the days before outsourcing was fashionable, doing everything yourself was seen as a good thing. Having full control over all everything made it easier to do integration testing and greatly reduced the combinatorial explosion of possible configurations. From an economic perspective, there may have been a payoff in locking your customers into buying all components from you.

Other manufacturers used the idea of "buy your disks from anyone" as a virtue.

This is not unique to the IT industry - for example, back in the 1960's the Rolls-Royce company needed automatic gear boxes for the Cloud series, which they purchased from GM in USA. The complete gear boxes were delivered to the factory in the UK, completely dissassembled and all the "unfinished" rough casting surfaces were machined to "RR Standards".

[Steven Schweda]

> Try again "Knowing nothing, I assume that
> there are..." is not an answer.

Sorry, but that's the best I can do with the
knowledge I have. I'd make up some
implausible bit of fiction, but I'm less
skilled at that than some other folks, and
I'd prefer to leave that job to an expert.

[Mark...]

Hi,

John, thank you - the information/point I think I was making all along. Yes I am aware of DSSI - wasn't that fun to configure! Best be careful though, moving off the topic again !!! Turbochannel & futurebus also went down the proprietry type route as alternatives to EISA/PCI.

Steven - just because YOU do believe that an answer is incorrect does not make it so. I think your idea of "because I don't know I'll just rubbish it anyway" is not in the spirit of the forum in which I believe most people are here to try to help people out - not muddy the waters.
As a side issue, I believe that they are looking for a new writer for Harry Potter books as the original author has had enough - maybe you should apply ??? :*)

Just to clear something up which Steven appears NOT to understand. I NEVER said the drives would not work, I just supplied some additional information as to why manufactures added extra value to there products - in this case disk drive f/w.

Either way the main point would be that the original noter [Dave] has resolved his question.

Mark...

[Steven Schweda]

> [...] I NEVER said [...]

And, so far as I can see, I never complained
about anything you said about disk drives. I
_did_ complain about that "non DEC memory did
not have the additional features" fairy tale,
which hasn't become any easier to believe in
the interval. But if you ever find any
credible evidence for it, please let me know,
and I'll be pleased to revise my opinion.