Richard Gambrell writes:
> Alfredo wrote:
> > Steve Dirickson b894 WestWin <[log in to unmask]> wrote:
> > >What about DCE? ...
> >
> > >If HP included the complete DCE product set on the 918DX (not just
> > >the RPC subsystem), we could build all sorts of interesting
> > >cross-system stuff using it.

> > Excellent suggestion.  I second Steve's proposal.

> Is DCE taking off for real?
> Is ODBC-32 enough?
> What about:
>   Remote methods for Java.
>   RPC (Cobra, etc.).
>   DCOM and more recent inventions.
>   etc.

There is a growing set of functionality available on the 3000 today,
although CSY loosening the screws on some proprietary items from the
9000 (source-wise) wouldn't hurt if we can get porting efforts down to
a level where things can truly be cross-compiled (that goes back to my
old original Proposition 3000 ideas).  There is plenty of development
on the HPUX side, but nobody seems to pose the question or face the
possibility that some of that code *just might* port to the 3000 for
little or no effort.

> Which "future standard architecture" is going to take off in the IT
> industry and which is worth the investment? What will the "next" SAP
> use? Which will the 3000 community adopt and use?

Regrettably people are looking increasingly at the "client" and paying
little attention to the "server".  And to play devil's advocate, many
of the servers can do these things more efficiently than the 3000 (web
cgi, Samba, DCE, ODBC, etc).  These things can help with co-existence
issues where the 3000 needs to remain for legacy system support, but
the justification is a bit weak (but getting much better) justifying
the 3000 as a more efficient, more economical option compared to other
solutions.

> As I've talked to many folks looking for staff lately, there is all
> together too obvious a trend: "GUI client/server or die!" There is an
> accompanying trend: migration from the HP3000 to something (NT, Unix,
> etc.).

With most attention directed to the client.  I feel that the 3000's
strength lies in:
   * ease of management,
   * ease of operation,
   * superior resiliency and reliability

Now if we have an ODBC-based application looking for a server, the most
economical solutions are WINfWG, WinNT, and/or Novell - cheap and run on
inexpensive hardware, but they aren't scaling well and are prone to the
usual weaknesses of a Wintel machinef (though granted they are getting
better).

Next we have the Unix server - exceedingly complex, lots of management
babysitting necessary, esoteric tools.  Worry about backups, fscks,
bootup, strange batch scheduling, cryptic command-line interface, etc.

Then we can look at "thick clients" - must a user really have a 200Mhz
32Mb pentium to run a client?  What's wrong with a telnet session or a
web form for "thin" clients or, heaven help us, tty devices?

> Can HP3000 software developers leverage the cheap popular PC tools,
> but obtain the quality, high performance, reliable result needed to
> complement MPE and Image/SQL? This is a tall order. Java looks to me
> like it might make it eventually. Bet on the right one and CSY wins,
> bet on the wrong one or spread too thin and CSY loses.

I would concur.  Naturally anything "value added" from Image/SQL
compounds the "Open System" options, but we have to acknowledge that
many "open" things aren't so "open" now -- Oracle development tools,
Cognos Axiant, Microsoft NT's proprietary add-ons, you name it.

But it looks like we're getting there.  It's still a bit of a challenge
to make a sale on a "new" application/3rd-party solution, but
nevertheless the 3000 is becoming a mature open system contender.

Jeff Kell <[log in to unmask]>