Re: OpenVMS Seminar in Toronto (2005-02-24) a few points

From: Bill Todd (billtodd_at_metrocast.net)
Date: 02/28/05 

Date: Sun, 27 Feb 2005 19:19:21 -0500

Robert Deininger wrote:
> In article <VKadnWo0jLjvk7_fRVn-iw@metrocastcablevision.com>, Bill Todd
> <billtodd@metrocast.net> wrote:
>
>
>>icerq4a@spray.se wrote:
>>
>>>Dave Weatherall wrote:
>>
>>...
>>
>>
>>>>I read your post and had exactly the same thought as Dave Frobie.
>>>>'He's said that before but I don't remember it being explained.'
>>>
>>>
>>>Explained (with as much information is available to the public) but
>>>possibly not understood,
>>
>>Actually, most people here seem reasonably bright and to be able to
>>understand just fine: it's the usual smattering of truth coupled with
>>the inevitable large serving of cHumPaq spin.
>>
>>HP has no motivation whatsoever to do anything but gouge Alpha
>>purchasers with whatever pricing the market will bear, and they do. Had
>>they been they motivated otherwise, they could easily have arranged, at
>>relatively little additional design cost, to allow small Alpha boxes to
>>share everything but their motherboards and processors with their Itanic
>>cousins.
>
>
> "Relatively little additional design cost"?!? Hogwash. In any case, all
> of the alpha systems, except the DS15, were essentially completely
> designed before the HP-Compaq merger. Before the merger, one group of
> engineers did not have access to the design specs from the other group.
> What you suggest would have been equivalent to a complete re-design of the
> Alpha systems after the merger.

Hogwash yourself: what I suggest would have been a relatively minor
redesign to package the existing board and processor into the supposedly
far-less-expensive Itanic enclosure and ancillary component
constellation. If carried out in conjunction with the design of the
first generation of the Itanic boxes after the merger was completed, it
would have added little to overall design cost.

But of course there was, as I said, no motivation to do so. If
anything, perhaps the opposite: the more attractive Alpha systems
looked, the more resistance to migrating to HP's platform of choice.

>
> DS15 was designed after the merger, but the overriding goal was fast time
> to market.

Why? What was the great rush to get the DS15/25 systems out, given that
they still used EV68 anyway?

   DS15 borrows most everything from DS10 and DS25, except for
> the power supply which is "almost" an XP1000 power supply. Borrowing from
> existing Alpha systems saved at least 6 months, likely more. The
> engineers already knew the older systems. They modified where necessary,
> with high confidence that most subsystems would work right the first time,
> and almost nothing would require more than 2 passes. Had they borrowed
> from the totally unfamiliar HP Itanium line, it would have taken much,
> much longer.

Given that EV6 DS10s could have offered roughly comparable performance
anyway, so what?

>
> There has been some fairly minor cross-pollination of server components
> since the merger. This was usually undertaken when there were obvious
> cost savings or supply chain simplifications to exploit. Even minor
> modifications cost man-months to design, qualify, and implement. Such
> changes would not be made for a $1 component; for a $1000 subsystem, it
> might be worth looking into.
>
>
>>Given that DS10 prices had dropped to about $5K at their low point (plus
>>OS license - e.g., you could buy a DS10 VMS system for well under $7K),
>>even without such further cost-reduction pricing wasn't *that* far above
>>the current crop of Itanics.
>
>
> Ah! Are you assuming the gross margins are about the same on the Alpha
> and Itanium families?

No: I'm assuming that if Compaq saw fit to sell the DS10 hardware for
around $5K HP could very reasonably have done so as well. And if Itanic
hardware really did cost that much less (whether due to higher expected
volume or to use of more standard components, or both), that HP could
have emphasized cost-reduction over time-to-market for DS15 and sold it
for even less.

   And why compare long-in-the-tooth DS10 with current
> Itaniums?

Because we were talking largely about entry-level systems here.

> Did you forget about DS15?

Covered above: please provide pointers to any benchmarks which show
DS15 with a sufficiently dramatic performance advantage over DS10 that
entry-level customers would have rejected the latter (especially at a
more attractive price) and been unwilling to wait for the former
(especially at a more attractive price).

By the way, what *does* a DS15 cost nowadays? And if it's significantly
more than around $5K (plus OS, if applicable), why is that?

   If you think the prices of alpha
> servers are artificially high, I guess some of your financial assumptions
> are very broken.

Your faith in your own understanding of the situation and in the
integrity of your employer may be personally gratifying, but it doesn't
really carry all that much weight out here in the wider world when
stacked up against past behavior, available evidence, and the eminently
obvious desire of HP to phase out one product in favor of the other.

...

> I've confined my observations to hardware I worked on as it was brought to
> market. (DS15, DS25, ES45, the Marvels, and all the Integrity systems
> currently supported by VMS.) I'm very familiar with most of the
> difficulties these systems encountered, and how the difficulties affected
> the ultimate schedule for shipping the products. Obviously the details
> are not public, and if HP wanted to make them public the info would not
> come from me.
>
> I never encountered a single case where any of these products was held
> back for non-technical reasons. (Marketeers will sometimes delay a
> product launch for a few weeks to align with their horoscopes or whatever;
> I consider a few weeks uninteresting and exclude such cases.) The bugs
> were fixed or worked around, and then the products were shipped.
>
> With perfect hindsight, every one of these products could have shipped
> faster, better, cheaper. So what? We move forward with the information
> we have day by day.

What a heart-felt defense. You probably even believe it yourself, just
as Fred probably did when he was describing the horrible obstacles which
the EV8 team would have had to overcome that would have pushed their
ship date out to God-knows-when - obstacles which those actually working
on that product stated were the product of Fred's fevered imagination
(or perhaps that of someone he had listened to a bit too uncritically).

Guess what? You really often don't 'know' nearly as much as you think
you do: your view is relatively local, and informed beyond that largely
by the words of people you may feel you have no reason to distrust but
that others have good reason to feel differently about.

>
> The atmosphere is generally frantic. Milestones always slip, deadlines
> always loom, and the entire team works like fanatics to get the damn thing
> out the door. Then a bunch of p****s in armchairs read a couple of glossy
> trade rags, some tea leaves, and a few chicken bones; based on their
> nanoseconds of real-word experience, they declare that the project was
> meant to fail and the mind-numbed slaves of Curly and Carly sabotaged it
> to keep their jobs.

Dear me, you sound so upset, and babbled on in that vein for a couple of
additional paragraphs which didn't seem worth quoting. But you really
ought to blame your employers for that rather than the messengers who
merely bring their manifest malfeasances to light.

- bill

Relevant Pages

  • Re: Signs of Deliberate Artifact?
    ... The argument from design was demolished more than 200 years ago: ... Here's Kant arguing against the teleological argument for the existence ... and human reason can never do without ... that is, the nature of different things could never spontaneously, by ...
    (talk.origins)
  • Re: OT - Cant we all agree????
    ... The reason is that so many decent, honorable, ... I've observed your thoughtful and measured responses to Powell over ... Jeff is who he is. ... Is he the most important design engineer in the ...
    (rec.music.classical.recordings)
  • Re: OT - Cant we all agree????
    ... posters, why is it that I keep wandering into threads which they have ... The reason is that so many decent, honorable, ... I've observed your thoughtful and measured responses to Powell over ... Is he the most important design engineer in the ...
    (rec.music.classical.recordings)
  • Re: Quad Core PC for less than a Mac Mini
    ... All the design work is done by the MB and case manufacturers. ... only a dual core on this page, no CPU for this motherboad can be ... Because *all* of the CPU options for this motherboard suck. ... So your answer is you're just bending over backwards to find a reason ...
    (comp.sys.mac.advocacy)
  • Re: Is OpenVMS certified yet for 64-way Hard partitions (GS1280 M64).
    ... I just thought I'd mention that with the last sale date for Alpha systems coming later this year, it is very reasonable for customers to evaluate their current system load and try to predict their future Alpha requirements. ... do you buy another M32 or do you inquire about an M32 to M64 upgrade with OpenVMS support. ... All the more reason to do evaluations/predictions today and pass them along to your account person/reseller. ... HP Pascal/MACRO for OpenVMS Project Leader ...
    (comp.os.vms)