Subject: | |
From: | |
Reply To: | |
Date: | Thu, 2 Jul 1998 08:17:31 -0700 |
Content-Type: | text/plain |
Parts/Attachments: |
|
|
Brian Williams writes:
[question snipped]
>On another subject: our HP3000 is to be replaced with an HP9000. Is
>there a listserver for HP-UX systems?
>[The Powers That Be are of the opinion that the 3000 will be gone in 18
>months: 4 years minimum is my estimate.]
HOGWASH. This is what 'Powers That Be' say when they either have their
own agenda, or they don't have a clue. You be the judge.
What I am going to say comes from public information that I have
acquired from HP presentations or random discussion over the last few
months. None of this information was acquired under non-disclosure.
1. HP3k sales are way up. In fact a recent visit from a ce to replace
a DAT was very informative. This ce works mostly on ux systems but
was helping out on the call because the assigned ce was busy. He
comments to me: "boy these 3000 systems sure have been selling
a lot lately".
2. Remember, HP has continually given us a system roadmap. At HPWorld, at
IPROF and at annual strategic product broadcasts. In all these
presentations there are 3k systems planned out for many years.
3. Guess which enterprise vendor is probably adopting PCI as a standard
system bus. You guessed it, HP. I saw a slide from a public presentation
with 3000,9000,nt systems all using the same bus (PCI). Now, why do
you think they would bother to migrate the 3000 to such a bus?
4. Maybe you didn't hear, but British Air just committed to buying 3k
systems for e-ticket applications. And remember, Southwest has been
committed to the 3k ever since they acquired Morris Air and found out
how great the box worked for the Morris applications. Southwest e-ticket
and soon (or maybe now) all reservations will be on the 3k.
5. I recently saw a slide in an HP presentation that suggested a major
platform introduction for the large/mid-range systems around 2002/2003
for mpe.
6. There are pa-risc chips still being designed/made that go way out in
performance. There will be 3k systems based on the 8500/8700/8900 chips.
7. HP is adding 64 bitness to mpe as I write this. Why would they bother if
the platform was going away?
8. Now that you have got this far, re-read #3. Sit back, think a minute, and
ask yourself this question: Why would HP convert all its enterprise systems
to PCI, a bus structure from Intel? Still hazy? then ask yourself this
64k question: What architectural advantage does HP gain by making this
transition? Still not getting it? Re-read #5. Still not getting it?
Then follow this history lesson:
a. HP had mpe on a custom cpu.
b. HP had unix systems also based on a custom cpu (several).
c. HP designed a new chip, PA-RISC and converted the mpe/ux worlds
to it and licensed the chip to others. The good step in considering
a standards approach to h/w.
d. HP adopts/promotes standard devices/interfaces (dat/scsi/etc).
e. HP/Intel design a new generation chip (ia-64) which Intel will use
for its high end systems and sell to other pc/server vendors. Given
Intel's market share, this new chip will be ubiquitous (ie: everywhere).
f. HP adopts the HP/Intel chip for its ux systems and of course will sell
Intel based nt systems with this chip in them.
As you can see, HP has been moving from a h/w architecture that was
extremely proprietary to a h/w platform that is very standard and
more commodity based. This makes too much sense.
g. The only thing left is for csy to base mpe systems on ia-64. And
that is precisely what I've been trying to tell you. I would bet
on it happening. When? Re-read #5.
I now return you to your regularly scheduled broadcast.
Duane Percox ([log in to unmask] v/650.372.0200x608 f/650.372.3386)
http://www.qss.com/ http://qwebs.qss.com/qwebs
http://qwebs.qss.com/faq3k
|
|
|