Re: OT: Sparc not dead yet



Doug Phillips wrote:
Main, Kerry wrote:

From: Doug Phillips


[snip ...]


And you think no one is working on those issues? At one time VMS had a
32 bit "bottleneck". x86 had all kinds of bottlenecks going all the
back to 4 bits. Today's x86 CPU's are more capable than both Linux and
Windows. That's the way evolution in this industry works. Hardware
advances, then software pushes its limits.


You also need to also remember that hardware and software gets better
with each release, but what many who think "x86 solutions will replace
everything" forget is that Customer RASS (reliability, availability,
scalability and security) requirements are also increasing
exponentially. Hence, while each new release of OS/HW gets better on all
platforms, the RASS bar is constantly being moved up as well.

Global DC consolidation, national and international regulatory and
compliance requirements, global 7x24x365 markets are causing many
Customers to re-evaluate how they deliver their IT solutions of the
future.



This is all true, Kerry. But how much money can x86's competition
afford to keep spending to stay ahead? The money being spent on x86's
evolution is huge. Sun is bleeding to death today and SPARC engineering
is a big bleeder. At some point, unless Sun is handed a miracle, the
SPARC division will probably be sold. Whatever is on the SPARC drawing
board now might or might not ever get to the market. Chips farther
along in the process probably will, and a return from that "shorter
term" market would be a reason Samsung or whoever might want to buy it.
Anyone willing to do a little homework should come to a similar
conclusion.

The problem I see with such reasoning is that it's looking only at one moment in time. Yes, right now the AMD Opteron is doing well. Do keep in mind that it's been called by some "little EV7". But whoever is on top today has no guarantee of staying there. Need I mention DEC, and even Intel?

Some other factors: Windows and Linux aren't the only OS's that could
run on x86. The focus on DT means that the majority of big users need
better distrubuted computing rather than huge monolithic crunchers.
There are only a few places where monolithic super-computing is needed,
and there will always be a place for the next "Cray". IBM is still
making money selling big-iron.

There will always be a place for the specialty systems, correct.

The Reliability-Availability-Security bar ("goal") is and always has
been at 100%. Scalability has a few different meanings depending on the
problem. As you say, IT users (customers) must continually re-evaluate
their IT service delivery. Likewise, IT solution providers must
continually re-evaluate how to best spend their R&D dollar.

Do you really want a world where everyone but one vendor just gives up? If you believe that you cannot win, and never start a race, then the opponent wins by default. Not everybody is a 'quitter'.

For many of the these Customers, I would suggest that QA/testing monthly
security patches (Linux/Windows) with their primary applications is
rapidly going to become unacceptable. And as we have seen by all of the
issues in the last few weeks, if Cust's do not test these applications
with these monthly security patches then they are open to all the issues
that hit the headlines. In a global market, a companies credibility can
take a serious node dive if this happens to them or worse, their Cust
data gets exposed because of one of these monthly security flaws.



This is all true, today. But, *any* system can be made secure or not
secure.

Well some have had lots of time, and aren't a bit closer, possibly further away.

VMS is unusual because its owners keep trying to reinvent the "wheel",
then they drop their wheel when they see the "old" wheels rolling on,
so they start over and try to invent a different wheel again. In the
meantime, the big wheels keep on rolling.



Can you expand on what you mean here?

Perhaps I missed an extract from an earlier thread, but this does not
make much sense on its own.



Substitute "CPU" for "wheel". Neither Alpha nor Itanium were
"revolutionary" technologies, they were attempts to advance and
redirect evolution. A true technology "revolution" will quickly
obsolete current technology. Money spent on Alpha and Itanium might
have been better spent elsewhere.

Who should work on evolution?

I think Alpha was a decent solution to the problems it was intended to address, and it did the job well. The only reason Alpha is not alive today is because Compaq did not want to be in the CPU business, and if they owned the Pentium, they'd still have killed it. Alpha is dead because the people in charge didn't understand the business they were in.

DEC ignored Win-tel until it was too late. Too often history does
repeat itself.

That I'll agree with.

What I won't agree with is the 'quitter' mentality.

Where would AMD be today if they just figured Intel would control the CPU world and it was no use competing? They could have concentrated on something else. Lots of money in electronics. But no, they believed in what they were doing, and they upset the hugh unbeatable Intel. Just read that AMD has a twenty something precent share in last quarter server shipments, up from 16 % or thereabouts.

--
David Froble Tel: 724-529-0450
Dave Froble Enterprises, Inc. E-Mail: davef@xxxxxxxxxxxxx
DFE Ultralights, Inc.
170 Grimplin Road
Vanderbilt, PA 15486
.