Mark Bixby writes:

> We just found out from our Oracle sales rep that Oracle won't be running on
>  MPE anymore beyond Oracle v7.3 (?).  Does anybody know more about this
>  termination of support for MPE by Oracle?  I've already told you all I
know,
>  and this tidbit probably filtered through at least 2 people before
reaching
> me.
>
>  And just to fan the flames of the Future Of The HP3000 debate, what *does*

>  it say about the HP3000's future when an industry leader like Oracle drops
>  support for it?

If the rumor is true, it probably says more about Oracle than the HP3000.

I remember very well about four to six years ago when Oracle announced that
the HP3000 would be its "preferred, recommended hardware platform" for its
Oracle database customers. My only source of doubt then was that HP had just
finished paying Oracle a ton of money (several million dollars) to port
Oracle to the HP3000. Indeed, one person from Oracle said at the time in one
of the trade magazines, that he "couldn't see how HP could ever get its money
back." Although he probably shouldn't have said what he did, he was almost
certainly right.

I and several others made mention often at the time that Oracle's ploy seemed
exceptionally cynical. Their loyalties would never lie with the HP3000.
Rather, selling Oracle on the HP3000 would be no more than a
spider-and-a-fly-like situation to them: attract new customers to Oracle from
the HP3000 installed base -- and then forgo upgrades (or at least be very
slow about it) and bad-mouth the HP3000 until they get these new customers to
move to whatever they consider to be the "standard" platform.

During those years since Oracle declared the HP3000 to be its preferred
platform, I've been interested to observe that Oracle has serially made
similar deals with other manufacturers. Just a short time ago, Oracle's most
recent preferred platform was DEC Alpha. I'm relatively sure that it will be
someone else here soon.

At the core of problem to Oracle is that it could never compete with a free
database -- especially one the quality of IMAGE, particularly so now given
its near-constant rate of improvement. And those who bought Oracle on the
HP3000 almost never did it to develop databases de novo but rather as part of
package, most often Oracle Financials. If you were to develop a new
application on the HP3000, you would be foolish to develop it in anything
other than IMAGE (or possibly, ALLBASE).

Oracle, in its heart of hearts, is a generalist. Adopting a generalist's
strategy means that you must move from platform to platform to survive,
essentially outrunning the competition from the specialists who can take
extreme advantage of the intracacies of whatever platform on which they've
specialized. To do this well, Oracle must spread as much FUD as real value.
"So that you don't get trapped," was a powerful message five years ago with
the advent of "open systems," before most people realized that adopting
Oracle was as much of a "trap" as any hardware choice.

Why wouldn't you want to develop new applications in IMAGE on the HP3000? The
question that is clearly on everyone's mind is: what is the long-term future
of the HP3000? How committed is HP to the HP3000?

My particular faith in the HP3000 as the best possible business computing
platform is absolute. My faith in HP itself is unfortunately somewhat less.
But these questions can be legitimately asked about every specific platform,
Windows NT included. The attraction of NT is not the hardware or software,
but the legendary committment of Bill Gates to his products. This common
queasiness about the future of any single platform is very much part and
parcel of Oracle's recipe for success. But it can't last forever.

In nature, species which adopt a generalist's strategy such as Oracle's only
survive well during those times of great disturbance, when a great deal of
new resource constantly becomes available. But once the world settles down --
and it always eventually does -- the generalist inevitably loses in every
different environment in which he originally prospered -- simply because
there comes to be a specialist that does the same task better and more
efficiently, at lower cost, than the generalist.

Most normally, under very stable conditions, generalists, if they survive at
all, are pushed off to small, resource-poor islands, too small for any
specialist to evolve. Or they become generalists themselves, as did Darwin's
finches on the Gallapagos, where a common ancestral form evolved to
specialize on a number of highly specific resource environments -- and in the
process, became species (meaning that they became distinct, non-interbreeding
lineages). Otherwise, in stable environments, the dominant biomass is never
made up of generalist species. Ultimately the world is always composed of a
number of specialist ("proprietary") species, because these are the
efficient, cost-effective solutions, each well-adapted and dominant in their
specific environments.

At the moment, the best strategy for the long-term success of the HP3000 lies
not in being either a pure generalist or a pure specialist, but in providing
a very well implemented SQL language shell for IMAGE (necessary for
client/server and application migration to the HP3000) and a high-speed ODBC
access method (necessary for JDBC and other network computer model
components), while retaining all of the HP3000's traditional simplicity and
high-speed intrinsic database access methods. This model is a mixture of
"emerging standards" (which, as Alfredo, has long pointed out, is perhaps the
ultimate oxymoron), a constant adaptation to trends ("investment
protection"), while retaining the extreme efficiency of those programs and
applications that wish to take advantage of it.

These are still interesting times, as the ancient Chinese curse has it, but,
as I've mentioned before, I do believe that the HP3000 is more like the
computer of the future than any other commercial computing platform currently
in existence. It is simple, tough-as-nails, and reliable enough to be
operated for years on end without a specifically trained data processing
staff. All it lacks is a bit of push.

Wirt Atmar