Re: HP World Magazine: For Business Continuity...the answer may be OpenVMS

From: Robert Deininger (rdeininger_at_mindspringdot.com)
Date: 05/11/04 

Date: Tue, 11 May 2004 10:15:15 GMT

In article <b8876400c5f6265491ad8f3db64c424a@news.teranews.com>, JF Mezei
<jfmezei.spamnot@teksavvy.com> wrote:

>Robert Deininger wrote:
>> I don't know exactly _why_, but it is the case. Digital built good
>> systems, but they were not very good at controlling costs.
>
>Aren't Alpha systems such as DS10s (and I would assume DS15s) built from
>industry standard parts ?

Partly.

>They use industry standard PCI bus, right ?
The proper punctuation, in this case, is "industry standard". PCI is a
nasty standard, with lots of wiggle room. Many hours have been wasted
getting PCI stuff to work right.

>Industry standard power supply,

No. Power supplies are mostly "custom", but ordered from a "menu" of
standard building blocks and features. Digital/Compaq rarely used the
same power supply twice in Alphaservers.

>drives, connectors, etc ?

Yes. But in the case of drives, remember that "industry standard" means
"lowest quality that is not guaranteed to cause riots". This can mean
lots of extra engineering time, which is hard to pay for on a low-volume
system.

...

>However, DS10s and DS15s are modern systems and under Compaq, their
>production costs should have gone down simply due to greater purchasing power
>of Compaq.

That sounds like Boardroom Boilerplate. Only marginally relevant for
low-volume stuff like Alphaservers.

>And in hindsight, it is clear that Compaq had no intentions of making Alpha
>succeed and as a result, was more likely to leave them priced uncompetitively
>and use accounting tricks to make their production costs look much worse than
>they could have been. Had Compaq made Alpha competitive and had the Alpha
>product line shown great growth, Compaq couldn't have done what it did on June
>25 2001.

Never let reality interfere with your fevered imagination. Nobody ever
needed "accounting tricks" to make Alphaserver production costs high.

>> I strongly suspect DS15 costs more to manufacture than the list selling
>> price of rx1600.
>
>OK, pray tell, from the system point of view, why would a low volume Alpha
>machine cost more to produce than a low volume IA64 box ?

Mostly lots of little things, I suspect. Power supplies. Fans. Better
re-use of lots of pesky little components in later generations of
systems. Fewer hours of labor to build the system. Better management of
the supply chain filled with thousands of individual components (some of
which are always gathering dust or going to the scrap heap).

A few big things. Alpha CPUs (bought from IBM) cost more than Itanium
CPUs (bought from Intel).

>> The only way to profitably sell Alphas much cheaper would be to cut the
>> cost to manufacture. That would require aggressively designing for low
>> cost (not a Digital strength)
>
>Both Compaq and HP are supposedly capable of producing low cost. When Compaq
>inherited Digital, it mentioned amny times that due to the combined purchasing
>power, it would be able to reduce parts costs for both DEC and Compaq
>products.

Very little of that happened for Alphaservers. Mostly Compaq disrupted
stuff that should have been left alone.

Look at an ES40 compared to an ES45. There's no fundamental change in the
way these systems are designed. Compaq's magic wand didn't really change
anything.

>Now, HP is even bigger and should have even lower costs for parts.

That formula works pretty well for paper clips. It's much hard to make it
work for complex system components with fairly rigorous specifications.

>Sorry, that excuse doesn't hold. Where there is a will, there is a way.

Whatever. I know better than to try to start my own company to build
computers. You are welcome to try if you like.

>Digital ceased to exist in 1999. Both Compaq and HP had plenty of time modify
>the DS10 and design the DS15 to change whatever resulted in higher
production costs.

DS10 was a Digital design. Compaq changed the logo. They didn't do a
ground-up redesign aimed at reducing the cost. Maybe they should have,
but it would have meant a long delay in getting the system to market. And
I don't think the Digital engineering culture could have done it without
radical upheaval. HP was far too late to alter the DS10 design.

DS15 is an HP product. It obviously borrows a lot from DS10, DS25, and
ES45. That was necessary because HP rushed it to market. Getting it out
the door quickly was more important than redesigning for low cost. So the
engineers (the old Digital server folks) did what they were used to doing,
and ended up with a typical Digital-priced system.

Relevant Pages

  • Re: HP World Magazine: For Business Continuity...the answer may be OpenVMS
    ... but they were not very good at controlling costs. ... Industry standard power supply, drives, connectors, etc? ... However, DS10s and DS15s are modern systems and under Compaq, their ... production costs should have gone down simply due to greater purchasing power ...
    (comp.os.vms)
  • Re: Industry standard is ridiculous!
    ... >> manufacturers that all had the same engine ... ... >> means no innovention, no progress ... ... >> next time you start singing the praises of industry standard ... ... > And the R&D costs for a new engine are probably on the order of the cost ...
    (comp.os.vms)
  • Re: VBA to name worksheet
    ... >>> descriptor name into the worksheet name. ... I have 4 pages tracking production costs on 4 widgets. ... >>> name, widget line, widget cost, etc). ...
    (microsoft.public.excel.programming)
  • Re: Future: 0603 versus 0402 parts
    ... The occasional ssop (0.65mm pitch) seems ... and to cover the investment costs of the production of smaller components). ... Mostly reliability concerns and ... production costs (especially hidden ones - ie change this resistor for this ...
    (sci.electronics.design)
  • Re: Future: 0603 versus 0402 parts
    ... The occasional ssop (0.65mm pitch) seems ... and to cover the investment costs of the production of smaller components). ... Mostly reliability concerns and ... production costs (especially hidden ones - ie change this resistor for this ...
    (sci.electronics.design)