• Subject: Re: Performance solutions through hardware
  • From: MacWheel99@xxxxxxx
  • Date: Mon, 19 Jul 1999 16:51:34 EDT

Frank wrote:

>  From:        fkolmann@revlon.com.au
  
>  Been on hols for a week, so reply is late.:(
>  
>  Mike Gillette wrote:
>  
>  > If in fact the AS/400 is source of your performance bottleneck, 
>  >it may be as much a workload mix issue as it is a horsepower issue.  
>  >The 9406/640 is primarily an interactive CPU 
>  >and suffers under heavy batch processing.  
>  >The new 7XX models and the older E-Custom Servers 
>  >are tuned to give the BPCS user an adjustable CPW mix between
>  > Batch and Interactive.  This has resulted
>  > in superior performance than what users have experienced with the 
>  >"Classic" AS/400 like your 640.
>  
>  Mike I am very interested in knowing what an Interactive CPU is as opposed 
> to a Batch CPU.
>  Our batch load is not great, or is my definition of BATCH vs INTERACTIVE
>  something which no longer applies to the new IBM computers.

I have recently been to several IBM classes & the instructors still define
INTERACTIVE = Human being interacting with the computer, whether twinax or PC 
is the work station
BATCH = the job is running without a work station in the picture, like off of 
a JOBQ

We have some jobs that have heavy updates & file i/o that should be run in 
different sub-systems like QBATCH from the user QINTER to utilize system 
resources productively.  This is an internal lapse in my data processing 
responsibilities.

The overhead needed to support these 2 modes, printer, communications, PC LAN 
etc. use different resources & various different models of AS/400 are 
designed to be extremely powerful in a particular area, so if you are 
primarily in another area, then you have the wrong box for your company.

>  I am perplexed as to how hardware can solve a problem that is inherently a
>  software issue.

There are problems that are difficult to diagnose was to what exactly is 
causing the performance problem.

>  To be precise,
>  if a programmer reads an entire file sequentially to update a record
>  rather than using a logical file to get to just the record that needs 
> updating, how will changing the CPU fix the problem.

In this example, the correct solution is through software, although hardware 
can make the extremely inefficient process get done sooner.

>  I am in the dark, I cannot understand how certain jobs take as long as 
they 
> do on the 9406/640.
>  Certain jobs just scream along, others seem to dissapear up a black hole, 
> and occasionally appear and to an update.
>  Call me paranoid, but this smacks of 'wait loops' to me, not that I would 
> ever suggest IBM would ever do anything like that 
> (does anyone remember a thing call a PCJX or Junior, never mind, its 
> just a red herring).
>  My first job was with a S3M15 so I' ve been about a while and I thought I 
> knew a thing or 2 but this stuff is like
>  the X-Files to me, or am I paranoid again.  

> Frank

Some jobs may appear to be in a wait condition if their access to system 
resources have a low priority.  Check out the rules in *JOBD for the JOBQ a 
batch is in.  This sort of topic is covered by IBM AS/400 classes in System 
Operator, Work Management, System Administration etc.

http://www.training.ibm.com/ibmedu/spotlight/as400.html = IBM's curriculum on 
AS/400 complexities

Al Macintyre

+---
| This is the BPCS Users Mailing List!
| To submit a new message, send your mail to BPCS-L@midrange.com.
| To subscribe to this list send email to BPCS-L-SUB@midrange.com.
| To unsubscribe from this list send email to BPCS-L-UNSUB@midrange.com.
| Questions should be directed to the list owner: dasmussen@aol.com
+---


This thread ...

Follow-Ups:

Follow On AppleNews
Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2019 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].