Re: How are CPU Speed and Overall CPW Related?

Let me give you some background, James.  I've been writing a lot RPG code
this past year - building a Web development framework.  And researching
whether a Web framework is a suitable alternative to 5250.

To help me optimize my RPG code, I'll sometimes write an equivalent in
Foxpro or Delphi and compare the difference.  It has puzzled me that my
$1,600 Laptop consistently out performs my $16,000 AS/400 - sometimes by a
factor of 10.  It further confused me after I learned that the AS/400 had a
200 Mhz processor - which I consider fast.  I wondered whether something was
constraining the AS/400 processor.  I have now learned that IBM has some
means of governing the CPU so that I can only get at about 20% of it's
capacity.  PC manufacturers don't do that.  The puzzle is solved.

Your explanation of the I/O support provided to 5250 applications may lead
someone to conclude that the AS/400 is not built to handle Web applications.
Perhaps, that's partly right.  I believe OS/400 is ideal for the Web.  But a
constrained CPU is not.  Web applications require lots of CPU because they
involve the handling of lengthy streams.  Apparently IBM is beginning to
realize that - the capacity of their new servers is a step in the right
direction.

Nathan.



> Date: Thu, 03 May 2001 14:14:15 -0700
> From: "James W. Kilgore" <eMail@James-W-Kilgore.com>
> Subject: Re: How are CPU Speed and Overall CPW Related?
>
> Nathan,
>
> As everyone has been trying to tell you, you are comparing apples to
> oranges by this measure.
>
> It does bring up the point that no single machine is best suited to
> solve all problems.
>
> It wouldn't take much to bring your 100mhz PC to it's knees with some
> other task running at the same time.  The AS/400 is not intended to be a
> single function machine, so it is not optimized for those types of
> functions.  And, yes, it will measure poorly when tested for something
> it was not designed for.  As any machine would test poorly for something
> it was not designed for.
>
> I'm not sure what you are trying to get at, but you must face that
> reality that the AS/400 is -not- the best machine at all things, But, it
> is the best machine when a single machine has to do all things all at
> once.
>
> Although I liked the motorcycle vs bus example, I'll give you a
> different one that actually relates to computers and the AS/400 in
> particular.  This has to do with the response degradation curve.  That
> is, under other systems, there is a direct, linear curve that
> corresponds with the number of users and their response time.  The more
> users, the slower the response time.
>
> IBM created an I/O subsystem that flattened that curve.  A single user
> was faster on the other system when compared to the AS/400, but 20+
> users were slower.  (The same was true in token ring vs ethernet)  Now
> why would IBM do this?  Well, from what I've read, IBM spends a whole
> bunch of money on researching things like psychological factors in user
> satisfaction.  What they found was that any given user was happier (more
> satisfied) with a system that provided a consistent response time than a
> system that went from fast, to slow, then back to fast again throughout
> the work day.  So if you get .5 sec response time as a single user, you
> will still get .5 sec response time with 20 or 100 or 10,000 users.
> (provided that you have the right model for the workload)
>
> I guess the point that I'm getting to is that the AS/400 never was, and
> never will be, optimized for a single user.  It is designed to be a
> multi user, multi tasking, consistent response time, back room, boring
> fixture.
>
> "Nathan M. Andelin" wrote:


+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---