I don't contribute to this newsgroup as often as I should.
But every once in a while, somebody says something that
prompts me to get up on my soap box.  Be forewarned, this
message is a little on the lengthy side...  :>
 
Terry Warns ([log in to unmask]) wrote:
: In marketing terms the 3k would be considered a cash cow.
: A cash cow is not bad for either HP or the 3k users.  It just
: means it will make a good profit for a long time.
 
Will all due respect Terry, I don't think this strategy is
very realistic.
 
The problem is that when you identify a product as a "cash
cow", then making a good profit quickly becomes the *only*
objective.  This means you cut all your costs (most
especially the the R&D budget for MPE) to an absolute
bare minimum; if HP had made the 3000 a cash cow, we would
today not have the POSIX environment on the HP 3000 - which
in turn would mean we would not have many of the new
technologies that are appearing on the platform now
(the developers of Appletalk, Oracle VII, Netware/iX, HT/TP
server and many others have all leveraged POSIX heavily, not to mention
the new applications that were mentioned in another thread).  If HP had
made the 3000 a cash cow, there would have been no reason to
bundle ARPA services.  Better to charge for it and maximize
profits.  Fundamentally, when a product becomes
a cash cow, customers *do* get gouged, because they wind up
continuing to pay for nothing more than what they already
have.  We aren't doing that.
 
: I believe that HP sales organization will have a difficult time
: selling an HP3000 today.  First , even though the 3k may be a
: superior solution, for many companies it is not THE solution
: that they are considering.  THE solution is UNIX.
 
I'm afraid you're dealing here with some old information.  A
year or two ago, I might have agreed with you.  After all,
if, in the customer's eyes, THE solution is UNIX, then all those
other solutions should just be relegated to the scrap heap.
Make 'em cash cows.  Y'know, HP almost believed that idea
just a few years ago.  We actually investigated the
possibility of moving HP 3000 customers over to UNIX
by mounting an MPE "shell" on top of a UNIX kernel; these
are the kinds of projects that come out of "cash cow"
thinking.
 
The idea never made it out of the investigation phase.  It never
became a project, or even a plan because we quickly realized
that UNIX doesn't have the reliability, the performance or
the integration of MPE - and a hybrid MPE-on-top-of-UNIX
system would be necessarily slower, down more often and
harder to get up again.  And when we talked to customes who
are using both 3000s and 9000s, again and again we were told
that 3000s were, and continue to be, faster transaction
processors, more reliable, and overall easier to manage than our own
UNIX systems.
 
If MPE had been a cash cow, we might have been tempted to
build the MPE-on-top-of-UNIX product and gradually moved all
the customers over to that model.  Why?  Because
it would have cut our costs - (one O/S kernel to support
instead of two, right?)  But we quickly realized that MPE
*shouldn't* be a cash cow, because it brings too many
characteristics to the table that UNIX does not.
 
Note that I'm *not* suggesting that MPE is THE solution
either.  If you want to run a good sized enterprise, then I
think you'd be silly to limit yourself to using *only* HP
3000s in that enterprise.  Similarly, I think you'd be silly
to limit yourself to using *only* IBM mainframes (a strategy
that may have once been viable, but now seems, well,
"quaint".)  Similarly, it's nuts to try to run a large
company on nothing but PCs, nothing but DEC VAXes, or for
that matter, nothing but UNIX boxes.  If you're looking for
one operating system, or one hardware platform, or one DBMS
that represents THE solution, then you're going to be looking for a long long
time.
 
To be honest, I think it's perfectly obvious that people who
think that THE solution is UNIX - (i.e., that they can solve
*all* their computing problems with UNIX) are naive.  UNIX
is not the future.  Neither is NT.  For the foreseeable
future, THE solution is going to be making heterogeneous shops work
- making networks talk to one another - making it possible
for applications to talk to multiple kinds of databases on
multiple kinds of machines using multiple architectures - using UNIX
where UNIX makes sense - using NT where NT (or Windows)
makes sense - and using MPE where MPE makes sense.  You
don't cut costs by moving everything over to UNIX.  You cut
costs by getting the most out of what you already have and
know how to use...
 
: The HP
: salesrep has got to compete for those sales and he has a very
: hot product to compete with.  With the UNIX requirement only,
: HP  salesrep can get in the door on many new accounts.  He will
: have no incentive to even suggest the 3k.
 
I can tell you that there fewer sales people today are
willing to go into an account claiming that they have the
"one true solution" to all the customer's problems, whether
that solution is UNIX, MPE or something that runs on a PC.
As for incentives, I cannot comment on your last statement in detail, but
will remind you that the incentives that HP sales reps have are not
entirely out of the control of Hewlett-Packard.   :>