>From: [log in to unmask] (Dan Hollis)
 
>MPE is dying. And will continue to lose market share until it does
>the following:
>
>1) Be viable on the desktop. There are desktop HP/UX machines. Why not MPE?
>Partially because of the terribly poor integration (minimal to none) with X
>windows. And partially because of the poor peripheral device support.
>
 
I would like to see this too.  If we boil MPE's woes down to the most
simplistic form,
we would (nearly) all agree that the biggest problem is coaxing more developers
and applications to the 3000, and diminishing support from HP.  There are other
problem, obviously, such as the one's Dan lists below.
 
>Linux, a free Unix, is available and can run on a 386SX/33 with 4mb RAM.
>With it, I can write applications with a GUI front end, that are portable and
>will compile and run on a high end Cray, AIX, DEC Alpha, Indy, etc. Nuff
>'said.
 
No, I don't think "nuff said"!  While I agree that portability is a wondrous
thing,
and much to be desired, it still has a patina of "buzz word" surrounding it.
At a fundamental level, I would say "So what?" regarding portability.  Let's
make
systems work.  Let's make applications work.  Let's go about making business
work.  To paraphrase someone (I forget who), "The business of data processing
is business!".  It's all well and good that we can sit in our ivory towers
looking into
our crystal balls (or Orthanc stones) prognosticating the demise of this and
the rise
of that, but the 3000 is still a damn good machine at getting transactions
processed,
monthends completed, orders processed, picklists filled, patients discharged,
and so on.
 
Don't misconstrue the above diatribe as saying that we shouldn't go down the
path
of openness, portability and so on.  We should.  At the same time, we should
continue to
make business cases which would woo HP back to the original love affair
(apologies to
Alfredo borrowing from his old theme regarding manuals) with the 3000.
 
>
>2) Scale beyond 16 CPU's. Right now a low end alpha can trash a high end
>3000 in OLTP by an order of magnitude or more, and yet end up costing 1/2
>what the 3000 does. It's pretty hard to sell machines that simply don't
>perform as well as their cheaper competetion. OS be damned, 3000's just
>don't have the "oomph" any more.
 
As a performance person, I agree that this must be addressed.  But it won't do
any good to have more than even the current amount of CPU's until Image/SQL is
no longer single threaded through the DBPUT/DBDELETE semaphore.  This
needs to be addressed even before the move beyond *8* CPU's.
 
>
>WHen a 3000 can run a massive web site like Alta Vista, I'll begin to take
>the 3000 seriously.
 
Not disrespectfully, I say "Who cares?".  It runs MicroWarehouse, Intuit,
Southwest
Airlines, lots of credit unions, mail order businesses, laboratory
information systems,
HMO's, insurance companies, manufacturers and so on.  The 3000 never was able
to be all things to all people.  My point is simply that it is enough to say
that is can be
all things to *some* people.
>
>3) Grow beyond the current 2gb file size limit. There are other limits that
>need to be eliminated as well. A 64 bit API is on the verge of being
>ratified, so Unix will be moving forward with 64 bits while MPE is left in
>the dust. (And don't fool yourself, MPE is no way no how 64 bits. The
>system calls dont support it in any way, and "64 bit pointers" really only
>address 32 bits.)
>
 
Again, as a performance person, I agree that this must be done.  Many people
voiced
this at IPROF, particularly in technical roundtable (or was it the
management roundtable?).
I don't think that MPE is a 64-bit machine.  I don't really hear anybody
saying it is.
 
I would disagree with Denys a bit and say that we do have the right to
demand certain things
from HP.  Just as HP has the right to say "no".  This is one of the things
that we should demand.
 
 
>Now call me skeptical, but I'm pretty sure that none of the above is going
>to happen.
>
 
OK, you're skeptical. :-)  I think that we have a very strong likelihood of
getting these things done.
Even though CSY appears smaller than before, a case can be made to say that
they are simply
"leaner and meaner".  Hopefully not too much meaner!
 
Time to put away my Orthanc stone and get back to the business of data
processing!
 
Bill
 
 
---
Bill Lancaster         Lancaster Consulting
(541)926-1542 (phone)  (541)917-0807 (fax)
[log in to unmask]       http://www.proaxis.com/~bill