Re: Thoughts on the book: DEC is dead, long live DEC

JF Mezei wrote:
Bill Todd wrote:
SCSI, if you'll remember, stood for "*small* computer system interface"
- and didn't even become an ANSI standard until 1986. The standard
remained 8 bits X 5 MHz and 8 devices (including initiator) until SCSI-2
came along in 1994 (doubling all three, which still seems a bit limited
for large VMS systems of that era).

My 1991 workstation (3100) has SCSI-2.

So after SCSI(-1) had actually become an ANSI standard, DEC not only embraced that but was willing to take a flyer on SCSI-2 even *before* it became an ANSI standard - hardly supporting your suggestion that DEC was trying to tie everyone into proprietary solutions.


The reason I chose DSSI as an example was that it was an example in the
1990s of continued development of proprietary stuff. Had DEC embraced
SCSI in the mid 1980s,

You mean as opposed to 1988, when it did start supporting SCSI on VAX? (Actually, http://www.hpl.hp.com/hpjournal/dtj/vol3num3/foreword.htm says 1989, but my impression is that DSSI and SCSI support appeared together, and DSSI support is dated 1988 in http://research.microsoft.com/~gbell/Digital/timeline/1988-5.htm )

it might have worked on elaborating the standard

IIRC it actually tried, but failed - again, taking the wind out of your argument (and since it was all wind anyway...).

in such a way that it could be used just as DSSI within a cluster.
(which later happened anyways).

No, it did not:

"For the early product life of MicroVAX and VAX Q-bus systems, the DIGITAL Storage Systems Interconnection (DSSI) bus was the storage choice typically available from DIGITAL. While the then-current SCSI-1 bus was about the same performance as the DSSI storage bus, DSSI provided dual- and triple-host support, SCS and MSCP capabilities, and a number of other capabilities well beyond SCSI. Accordingly, SCSI connectivity was a comparatively late feature in Q-bus configurations. (Some of the capabilities found in DSSI still do not exist in SCSI. Ever SET HOST into a disk controller to directly adjust its settings?)"

http://hoffmanlabs.org/openvms/hwvax/hwqbus.shtml (!)

In fact, even now SCSI is supported only for storage (rather than host <-> host) VMS cluster use - see page 3 of http://h18000.www1.hp.com/info/SP2978/SP2978PF.PDF

While at some point SCSI did grow to support dual (and possibly even more) initiators, ISTR that Tru64 discovered some of these SCSI limitations when it was starting to investigate fail-over support, though it managed to kludge its way around some of them.

....

impossible and usually disastrous goal for an engineering firm) and
seeking proprietary advantage do you find difficult to understand?

And thsi is where DEC really failed the paradigm shift.

Stop babbling and pay attention for once, JF: the point above was that you completely misrepresented one kind of behavior (trying to tie people into proprietary technology, although even that assertion is becoming increasingly dubious since DEC started supporting SCSI soon after it had been accepted as an ANSI standard) for another (trying to attain perfection).

....

you'd have noted that DEC didn't suffer due to added markup, it suffered
because *the nature of the playing field changed* due to *the
introduction of new, disruptive technology*.

That disruptive technology is called COMPETITION.

But it's a *qualitatively different form* of competition than the simple mark-up practices that you stated were responsible, idiot: that new competition could have had mark-ups exceeding DEC's and still shouldered its way in because its *costs* were so much lower.

....

The same is happening with clustering now. The clustering offered by
Unix and even Windows is more than enough for many people who cannot
justify paying a premium for VMS clustering even though it is by far
still superior.

Horse***:

1. VMS clustering is by no means 'far superior' in many contexts (in some ways it's still at least somewhat superior, but what you pay in terms of committing to a niche platform that may be on its way out may far outweigh that).

2. VMS isn't priced to compete because it's a cash cow, not because it would be impossible to price it to compete.

....

*Alpha* is nothing approaching 'commodity hardware', nitwit.

Insults aside, Alpha could have been commodity.

*BUT IT WASN'T*, idiot. Your argument was not hypothetical in some never-never land, it was about current reality - and that reality includes the fact that Alpha is nothing like a commodity product with commodity costs and pricing.

....

You still don't get it. DEC *didn't need to change from its roots*: it
just had to understand where it should compete and where it shouldn't
try to, rather than assume that it could (and should) be all things to
all customers (something it actually never *had* tried to be until it
grew so large that it seemed to think this had become a requirement).


Sorry, this is the very mentality that brought DEC down and the very
mentality that is currently hindering VMS.

You really are a fucking idiot, JF. Fortunately, most people recognize this and ignore you, but every once in a while it's worth letting newcomers in on the secret.

If you feel that you can
retreat to your own small niche market of customers still willing to buy
your products,

It's not about retreating to a niche, moron: it's about understanding what your strengths and weaknesses are and trying to capitalize on the former while minimizing the negative impact of the latter.

DEC didn't have to *produce* PCs to make a lot of money off them: it just had to figure out a way to *leverage* them.

you have to expect your competitors to eventually invade
your niche and kill you.

It's been another 20+ years now and that 'niche' is still very lucrative, by the way (not that I was suggesting that DEC confine itself there, as explained just above).

....

unique strengths like Rdb? What, exactly, did Palmer ever do other than
slash & burn? Where did he ever try to *create* opportunity?

In the first 2 years of his helm, Palmer did make big changes to Digital
that were not yet the "slash and burn".

Such as?

....

during
his first couple of years, he fixed up many things internally
(accounting etc).

Assuming for the moment that accounting was a real problem, that's one.

And he brought in new blood to try to change the
cultute to be more profit oriented.

Horse***. The culture was *already* too 'profit oriented', and adding another dose of 'MBA disease' wasn't going to help. It needed to get back to being *customer* oriented and *competency* oriented.

- bill
.