Ok, I'm finnaly breaking down and replying to the latest doom of the
3000 thread...
Jim writes:
>Since there presently is such a dearth of object languages for HP3000,[snip]
Gee Jim, you have Python (which is almost everything you could want in an
object language), and Mark's g++ port (I must not have been too good
at convincing everyone how awesome Python is, since there have only been
two downloads so far (and one of those was from Finland)).
> #define SOAPBOX ON
Ahhh, another Unix/C'ism'. Tsk, tsk. Scary how it creeps in :-)
DEFINE SOAPBOXMODE = TRUE# << BEGIN PONTIFICATION >>
The ImminentDeathOfTheHP3000 seems to be getting to a lot of people again
the last couple of days. Here are a few random opinions (and some
annoying metaphores)...
First of all, we need to understand how the Computer System Circle-of-Life
works:
Software Developers
^
/ \
/ \
/ \
<------->
HP Customers
1) HP convinces the Software Developers that the HP3000 is a good platform
to *develop* *new* applications on.
2) The Software Developers provide new solutions on the HP3000 that
customers want to buy which are only available on the 3000, or run
substantially better on the 3000 than on other platforms.
3) The customers buy the applications and the HP3000s to run them on.
4) The cycle repeats.
Each step justifies the previous one and provokes the following one.
This is how it has always worked. Like a three legged chair, it will
fall over if you remove any one of the three parts. No two of the three
groups together can save the 3000.
Even if the end users and HP both agree to do *everything* they can to
support the 3000, without new applications it's not going to survive
forever.
THE PAST
Let's look at how the 3000 became mortally wounded in the first place.
Some years ago all there was were proprietary systems from IBM, DEC, HP,
and others. The HP3000 had a small piece of this (most people don't
seem to remember *how* small the 3000 business has always been relative to
the rest of the industry and seem to feel that if the 3000 can't compete
with everything else, it isn't worth bothering with). DEC always sold
more Vaxen than HP sold 3000s, and IBM has sold more AS400s in a month
than the total number of 3000s HP had *ever* sold.
Along comes Unix. Unix started out as an operating system that nobody
was running, but everybody was talking about. In Computerworld, Infoworld,
etc., all you heard about was how Unix was the future. The reasons Unix
was touted as the future were A) It was new, and B) it was "Open". Let's
look at each of these.
First it was New. It was the "latest thing". People are always
optimistic that the next generation of product will solve all the
ugly problems with the current one. For most of the decision makers
who choose operating systems, the only things they remember about
their current system are its problems. The only things you hear
about a new technology that nobody is using yet tend to be the good
things. Worst are the consultants and writers who act as zealots
for the new one-true-operating-system and tell everyone how they will
be damned if they don't immediately convert. Of course by the time
anyone does, the zealots have found the one-truer-operating-system
and are back trying to convince everyone to move on to it. These
people never stick around long enough to be there when the bits hit
the fan.
Second it was "Open". In the future, software written to run on
Unix would run on any system, regardless of manufacturer, as long
as that system was running Unix. At least this is what we were
told. As a software developer, how could you ignore this? First
it meant that you could no longer be subject to the whims and
future of a single hardware manufacturer, and second that if your
competitor ran on Unix and you didn't, you could be in big trouble.
So all the software developers started looking into Unix.
It's very hard to port from a proprietary platform to unix, so most
found that they were facing either a massive porting effort, or the
development of a new product entirely. Pretty soon all the new
products are being developed under Unix. Oops.
Most customers are not in a position to know which operating system
is the best for them, but the software developers support the platforms
that the customers ask for, and everyone was either asking for Unix or
asking if the developer was going to support Unix in the future.
At this point the sucking power of Unix was so strong that everything
started heading that way. HP didn't help matters by promoting Unix
over MPE for a number of years, but really, the current state of affairs
is not as much HP's fault as a lot of people here like to think it is.
Unix killed the Vax and VMS off so thorougly that MPE now gets much more
press than Open VMS does. DEC suffered because they failed to embrace
Unix early enough and tried to stick with the proprietary way of doing
things. If HP had tried to do the same thing, I think the 3000 would
be *worse* off than it is today.
THE PRESENT
Today the future for the 3000 does not look very bright. The applications
just aren't there to provide the growth that it needs. There are only a
couple "best-of-breed" applications left on the 3000 that could sell new
applications. Whatever happened to ASK anyway? Seems like five years
ago half of the customers I talked to were using MANMAN. Now I haven't
heard a peep out of a MANMAN customer in ages.
Porting applications to the 3000 using Posix can help those who already
have 3000s keep them, but who would buy a 3000 to run a Unix application?
Sure it might be more reliable, maybe even faster, but when you have a
problem and call the software developer for support are you going to get
to talk to someone who understands the 3000? To thrive, the 3000 needs
3000 applications, not Unix applications.
THE FUTURE
Ok, can the 3000 be saved? Perhaps. Will it happen? Mostly it's now a
question of commitment on HP's part. Let's look at what it would take
for the 3000 not only to survive, but to grow into something that the
Unix and NT people have respect for (they don't have to like it, just
respect it :-)
The Circle of Life needs to be reestablished. There need to be new 3000
applications to sell 3000s. HP needs to create an environment that will
attract developers to the 3000. Some random ideas for what HP might choose
to do if they decided to do this (as if anyone cares what I think):
1) Recognize that they have billions of dollars invested in the HP3000
program and that currently their investment is rapidly evaporating.
Once it is allowed to deteriorate, it *cannot* be gotten back.
2) Decide that the 3000 is worth keeping around even if it won't ever
sell as many boxes as Unix or NT. The 3000 was always a small part
of the over all minicomputer market and there's no reason why it
can't profitably continue this way. It also acts as a differentiator
between HP and other vendors. They have something that no one else
has.
3) Communicate these decisions internally. Currently most people inside
HP don't seem to think the 3000 has a future, and this comes through
when they talk to people. It is also reflected in the fact that
much of the rest of the company has not heard of the 3000, and if they
have, they feel that *their* product is a replacement for the 3000.
An HP employee (not connected with CSY) told me the other day: "This
product is dying. It is going away." If that's the message HP internal
people are getting, what about the rest of the world?
4) Communicate to developers a commitment to support the 3000 for the
foreseeable future (ok, so that's only two weeks in this business),
something like the next ten to fifteen years would be nice. This
is the most important thing for HP to do and it's the least likely.
But the rats won't come back until the sinking ship stops leaking.
Create a small group of evangelists to visit developers of vertical
market software and explain why they should think about developing
their next generation product on the 3000. Vertical markets are key.
Let HP-UX and NT have the everything-to-everybody horizontal market
world. Let the 3000 do what it does best: be a platform for development
of world-class solutions to vertical market problems.
5) Downsize, downsize, downsize. Eliminate today's CSY organization
and replace it with a software only group to maintain and support
MPE. Organizationally merge this group with the Response Center's
MPE group. The new organization would get all the software support
revenue, as well as money from new MPE sales. This group must be
empowered and supported sufficiently to enable them to keep up with
their Unix and NT counterparts in supporting new HP CPUs and
peripherals. This means commiting to moving MPE onto the Intel/HP
architecture and doing a good job of it. At the same time:
6) Divorce MPE from the 3000. This means that there should no longer be
HP9000 and HP3000 products. Just HP PA-Risc systems, most of which
run HP-UX (or maybe NT), and some of which run MPE/iX. Sell the
right hardwware to the customer (PA Risc or whatever follows) for
the right hardware reasons, and sell them whichever operating system
they want to buy. You have one sales force selling one type of system.
You have one hardware development group and one marketing group. All
we ask is that the Marketing group tell people about MPE in the same
breath they tell them about Unix. They don't even have to push it. As
soon as the customer says they want HP-UX (because that's what their
software supplier told them they need), HP doesn't need to mention MPE
again. But if someone wants MPE, there should be no reason for them
to try to steer the user to something else (in fact, MPE will probably
cost more than Unix or NT, so there's a chance the salesperson would
make more money selling MPE).
If you did all this then you might have an environment that could attract
a few developers. All it would take is three or four really good
applications to sell as many boxes with MPE as ever. HP would work with
each developer to help sell their software without regard to which OS
ir runs on. It's as though the OS was coming from the software developer
and HP was just selling the hardware as far as the HP sales force would be
concerned. The trimmed-down software only CSY should be able to support
themselves with support and upgrade dollars until there start being new
customers (it would take at least a year or two for any of this to result
in new business).
Everything in life is cyclic. We still haven't reached the peak of the
Client-Sever-Destributed application phase, but before long people will
realize that there are as many trade-offs in that kind of application
architecture as any other. I'm not suggesting that MPE limit its developers
to traditional MPE style applications (View/COBOL), but there may soon
be a market for simple, cheap, and reliable solutions to real world problems
again. Does an order entry clerk really need a 16Million color display?
Do you really want the number of System Managers in your organization
equal to the number of Users because each one has a 'PC' as powerful and
complex to manage as a 3000?
In Douglas Adams's 'The Hitchiker's Guide to the Galaxy' there's a point
at which they talk about all the problems with the complexity of the
elevators designed in the future, and mention someone who rediscovered and
patented a device he saw in a history book, and thus became fabulously rich.
That device was the staircase.
G.
END.
|
