"Green Screen Tax" - How to avoid it?
Frank Whittemore
advantage of IBM's Operation Greensteak promotion. In the process, my
green-screen tax rate would appear to have increased!
At some point in time, it would seem that my shop must do something to
avoid this tax.
My first question is - what are traditional mostly custom written RPG
shops actually doing and planning to do to avoid the green screen tax?
Next question - how would you calculate my shop's tax rate? Here is
some raw data to ponder. Our old AS/400 had CPW ratings of 220/30.
Dividing the 30 by 220 indicates that the governor would kick in at
about 14% Interactive utilization. Our new iSeries has CPW ratings of
1070/50. Dividing the 50 by 1070 indicates that the governor will kick
in at about 5% Interactive utilization.
Final question - how should I explain the following words found in the
latest V5R1 cumulative PTF instructions to the ower of the company?
"Vital PTF information: Application of PTFs within this cumulative PTF
package may disable or render ineffective programs that use system
memory addresses not generated by the IBM translator, including
programs that circumvent control technology designed to limit
interactive capacity to purchased levels. These PTFs may be
prerequisites for future PTFs. By applying these PTFs you authorize
and agree to the forgoing."
Dan Hicks
> Final question - how should I explain the following words found in the
> latest V5R1 cumulative PTF instructions to the ower of the company?
> "Vital PTF information: Application of PTFs within this cumulative PTF
> package may disable or render ineffective programs that use system
> memory addresses not generated by the IBM translator, including
> programs that circumvent control technology designed to limit
> interactive capacity to purchased levels. These PTFs may be
> prerequisites for future PTFs. By applying these PTFs you authorize
> and agree to the forgoing."
It's essentially saying that it will disable "Fast/400" or
whatever it's called.
--
Dan Hicks
The idea that is not dangerous is not worthy of being called an
idea at
all.
Michael Anderson
"Frank Whittemore" <fra...@adelphia.net> wrote in message
news:bcdcd347.02093...@posting.google.com...
> My shop just migrated from an AS/400 170 to an iSeries 270 taking
> advantage of IBM's Operation Greensteak promotion. In the process, my
> green-screen tax rate would appear to have increased!
>
> At some point in time, it would seem that my shop must do something to
> avoid this tax.
>
> My first question is - what are traditional mostly custom written RPG
> shops actually doing and planning to do to avoid the green screen tax?
>
> Next question - how would you calculate my shop's tax rate? Here is
> some raw data to ponder. Our old AS/400 had CPW ratings of 220/30.
> Dividing the 30 by 220 indicates that the governor would kick in at
> about 14% Interactive utilization. Our new iSeries has CPW ratings of
> 1070/50. Dividing the 50 by 1070 indicates that the governor will kick
> in at about 5% Interactive utilization.
Where do you come up with dividing the batch CPW by the interactive CPW?
Your 170 had 30 CPW of interactive and now you have 50, what's the problem?
-Mike
Walter
> latest V5R1 cumulative PTF instructions to the ower of the company?
> "Vital PTF information: Application of PTFs within this cumulative PTF
> package may disable or render ineffective programs that use system
> memory addresses not generated by the IBM translator, including
> programs that circumvent control technology designed to limit
> interactive capacity to purchased levels. These PTFs may be
> prerequisites for future PTFs. By applying these PTFs you authorize
> and agree to the forgoing."
There is a program called Fast/400 which removes IBM's
get-rid-of-AS/400-customers interactive break.
So IBM built a program into the cum PTF which disables Fast/400.
So expect the next Fast/400 update which works also when this PTF is
applied.
Walter
Duce
What the heck is the Green Screen tax?
At first, our first thought was to simply upgrade our disk size, but for the
right price we would go with the GS Promo.
However, if there are hidden costs, I would like to know of them. I noticed
that our P05 tier would go to P10 and that could be a deal breaker.
-Duce
."
Bradley V. Stone
where your green screen apps do most of their processing. There are
ways around it, but they all cost money and time. Most newer iSeries
are tuned for batch processing. Just another way that IBM is trying
to force you into Java, Websphere, etc. Which IMHO is pure crap. You
should get CPU, and be able to put it where you want, interactive or
batch. It's a big scam IBM is using to force you to do things the way
they think is best, and not to mention will make them more money in
the long run.
There are some gotchas with the GreenStreak upgrade and I'd encourage
anyone considering it to read Al Barsa's article in the latest issue
of iSeries Experts Journal. Some of the key "gotchas" are that sure,
you're paying less for a machine, but you're also moving up in
processor level so support and software (IBM and 3rd party) will cost
more for upgrades.
Brad
www.bvstools.com
On Tue, 1 Oct 2002 07:54:48 -0700, "Duce" <Quadr...@hotmail.com>
wrote:
MBBrutman
I don't want to get into the 'green screen tax' debate - everybody knows
it all already. However ...
If you have access to the RPG, COBOL, or the other code making use of
the green screen, then you can restructure it to use less interactive
CPU. The idea is to keep the user interace on the green screen, if that
is what you desire, and move the 'heavy lifting' to batch jobs. The
green screen part of the application can ship it's work to the batch
part using data queues, database tables, sockets, whatever.
This obviously isn't a weekend job - it requires restructure and
rewrite. However if it is done well, your presentation code (green
screen code) will be separate from your business logic (batch code).
This makes it usable not just for the green screen users, but also for
users who can have a special client on a PC, web users, etc. (That
batch code can be reused now that it doesn't have the screen control
logic in it.)
Having the batch jobs also fits IBM's pricing model for the system -
cheap batch power. It takes very little interactive CPU to drive
screens and ship the data to batch processes, which should keep you
under your interactive cap.
Mike
Rich Knech
The age of circumvention has begun on the AS/400. This has been going
on in the PC world for years.
Anyone got a good AS/400 Hex editor or Disassembler? :^)
The folks that bring you Fast/400 will -HAVE- to get the latest PTF's
as soon as they come out to test them to see if there is a "break
Fast/400" PTF in it so they can get a patch out right away to their
customers. Of course IBM could always use the DMCA against them, just
like their counterparts in other industries are doing to others. I
would expect something like this to happen.
"Walter" <wal...@nospam.com> wrote in message news:<zuam9.365$ro2....@news.uk.colt.net>...
Mark Ford
We upgraded (for batch performance reasons) from a 720/2063/1503 (810/240
CPW) to an 820/2397/1524 (2000/240 CPW) - both with same interactive CPW -
and were soon quickly impacted by CFINT despite no increase in interactive
workload demand. In percentage terms, this is a reduction in interactive CPW
capacity from 29.6% to 12%.
Surely if a user is running an intensive function which had, say, a 25/75
CPU/IO resource demand split, then wouldn't they would be demanding 25% of
the CPU resource irrespective of the CPW rating (just completing faster with
a quicker processor).
Our solution was to play ball and upgrade the interactive feature to give us
560 interactive CPW (or 28%) and strangely everything was OK again!
Mark.
"Michael Anderson" <michael.p...@onemain.com> wrote in message
news:Uy7m9.2139$OB5.1...@newsread2.prod.itd.earthlink.net...
kra
> to have increased!
The price of interactive capacity actually went down, significantly.
How did enormous discounts in batch capacity get translated into
"interactive tax" anyway?
> My first question is - what are traditional mostly custom written RPG
> shops actually doing and planning to do to avoid the green screen tax?
Buying software to disable IBM's interactive governor is not good
strategy because IBM is committed to blocking that temporary
workaround.
> Our new iSeries has CPW ratings of 1070/50. Dividing the 50 by 1070
> indicates that the governor will kick in at about 5% Interactive
> utilization.
By running a CPU bound benchmark I found that a model 820 (1100 total
CPW) has a CPU speed that is 26 times greater than my model 170 (73
total CPW), for example. 5% of a faster CPU may actually be better
than 30% of a slower CPU.
To lower your cost of computing, the best long term solution is to
move workload to batch, and avoid linking a job to a 5250 device. Use
HTML and HTTP rather than the 5250 data stream and Telnet, for
example.
Nathan M. Andelin
www.relational-data.com
Frank Whittemore
----------------------------
Mike -
The problem is that I can only use about 5% of my overall iSeries
model 270 capacity for Interactive processing!
The problem is how to effectively use more of my machine's capacity?
The problem is PCs, a PC LAN and an outsourced web site running on
Wintel are steadily invading what was at one point in time an 100% IBM
midrange shop!
Am I clear yet?
My real question is what are you and others doing to more fully
utilize the capacity of your AS/400 and iSeries machines by learning
programming techniques or whatever to get around that annoying,
expensive governor imposed by IBM?
Please help educate me as to what traditional, custom written, RPG,
green-screen shops are really doing today to fully utilize their
systems and to keep the IBM imposed governor from driving them right
into the hands of the waiting Wintel folks!
Michael Anderson
I think you took my reply a little too far. I understand that out of the
total CPW, you have a less percentage allocated to interative jobs, but
don't you still have more horsepower for your interactive jobs than you did
on your 170?
As I understand the CPW figures, the first figure is the total CPW for the
machine and interactive gets x amount out of that. Please correct me if I'm
wrong...I don't claim to be right.
I'm not doing anything....we're still developing at V3R2 with no plans to go
to a RISC environment for 3+ years. We use a non-Y2K compliant version of
SSA's AS/SET to generate RPG code for the lazies who don't want to learn
RPG....ILE and RPG IV are a fantasy.
I can't anwser your last question. We are a developer of AS/400 software
for the market place, not for internal use. Our significant competitors are
not on a Wintel platform.
Maybe you should try the used market. I saw a 640 on eBay for under $4K.
About 320 CPW...no nasty governor and still upgradable to V5R2.
-Mike
kra
My 1st response was a little muddled. Let me try again.
I use a generic SPECIAL device instead of a WORKSTN device, which
routes display file record and indicator buffers to custom exit-point
procedures which interface with the HTTP Server (instead of Telnet),
and generate HTML (instead of a 5250 data stream).
An HTML stream is larger than a 5250 stream, which requires a little
more CPU time to generate. But changing your RPG programs to use Web
technologies is a good way to avoid IBM's interactive governor.
For example:
http://www.relational-data.com/rdweb/rd1200s/green.html
shows screen shots of a converted RPG program.
You're welcome to try the Web version at:
http://www.relational-data.com/rdweb/gotomenu.html
Note that Internet Explorer is required because I used a few IE
specific features in an attempt to emulate the 5250 keyboard. Feel
free to browse, add, modify, or delete records. The data is just
nonsensical test data.
Frank Whittemore
> I think you took my reply a little too far. I understand that out of the
> total CPW, you have a less percentage allocated to interative jobs, but
> don't you still have more horsepower for your interactive jobs than you did
> on your 170?
Yes, we do have 67% more Interactive horsepower. 50 minus 30 = 20 more
Interactive CPW. 20 divided bt 30 = 67%. I should point out that our
users love this upgrade. Work flies through the new faster iSeries 270
- especially the batch work that we do have.
>
> As I understand the CPW figures, the first figure is the total CPW for the
> machine and interactive gets x amount out of that. Please correct me if I'm
> wrong...I don't claim to be right.
You are right. Divide the total CPW by the Interactive CPW and the
resulting CPU percentage is where the governor kicks in if you are
only running Interactive work. My shop encourages users to run
queries. It only takes a couple of queries running at the same time to
cause the governor to kick in. As a result, we force our users to run
large queries in batch.
>
> I'm not doing anything....we're still developing at V3R2 with no plans to go
> to a RISC environment for 3+ years. We use a non-Y2K compliant version of
> SSA's AS/SET to generate RPG code for the lazies who don't want to learn
> RPG....ILE and RPG IV are a fantasy.
>
> I can't anwser your last question. We are a developer of AS/400 software
> for the market place, not for internal use. Our significant competitors are
> not on a Wintel platform.
Wow! What you say here leads me to wonder how long it will be before a
competitor does come along and start eating your lunch or should I say
into your market share?
>
> Maybe you should try the used market. I saw a 640 on eBay for under $4K.
> About 320 CPW...no nasty governor and still upgradable to V5R2.
Thanks for taking the time to answer my questions.
>
> -Mike
Philip Jamison
> Frank,
>
> My 1st response was a little muddled. Let me try again.
>
> I use a generic SPECIAL device instead of a WORKSTN device, which
> routes display file record and indicator buffers to custom exit-point
> procedures which interface with the HTTP Server (instead of Telnet),
> and generate HTML (instead of a 5250 data stream).
>
> An HTML stream is larger than a 5250 stream, which requires a little
> more CPU time to generate. But changing your RPG programs to use Web
> technologies is a good way to avoid IBM's interactive governor.
>
>
Hi,
Having read this and other posts, I would like to try to change a big
Order Enquiry pgm (will kills our 170 )to shift the I/O to batch. How do
I go about this (using only DDS,CL and RPG (3 or IV) ie no HTML or true
ILE etc . There are many posts mentioning the theory but no detail. What
I need to know, amongst other things, is:
1) Does the batch job run all the time or just for the job ie submitted
from CL and ended on exit? If all the time, is there one per program?
2) Does it use Dataqs 'connected' to the display file?
3) What if the screen job dies, is the batch still running?
With respect to this and the HTML-style displays,eRPG etc I know
Interactive coding and batch jobs with dataq's etc, I just can't get to
grips with connecting the two on an ongoing basis.
Any help and/or links much appreciated.
Regards,
Philip
AS/400 Model 170, V4.4
Karl Hanson
(Apologies in advance for the long-ish post)
You could think of a data queue as a pipe between 1-n client
(interactive) and 1-n server (batch) jobs. Each server job could have
logic something like:
LOOP:
-- Receive transaction from input data queue
and specify a wait time.
-- If receive timed out (no transaction)
Check if this job should end (eg RTVJOBA ENDSTS etc).
If so, exit LOOP
-- If transaction received
Call program (eg RPG) to process it.
Send result to output data queue.
-- Iterate LOOP to receive next transaction.
The client job logic could be something like:
LOOP:
-- Get transaction data from green screen, eg RPG EXFMT.
-- If user wants to quit, exit LOOP
-- Send transaction data to server input data queue.
-- Receive result from server output data queue.
-- Display result on green screen.
-- Iterate LOOP to get another transaction.
Note the SENDERID data queue attribute - see CRTDTAQ command. If
serving multiple clients with a server job, this can be useful to
ensure transaction results are routed to the correct client. The
server input data queue could be created with this attribute, so each
transaction received automatically comes with the client's unique job
name (sender ID). The server output data queue could then be keyed by
(entries contain) the original client's job name, so the client could
receive (by key) only result messages for transactions it sends to the
server. In this way one or more server jobs (daemons in Unix lingo)
could stay running, waiting for transactions and serving multiple
interactive jobs.
Alternatively a SBMJOB could be done so each interactive job has a
"shadow" batch job, with unique/dedicated data queues for each pair.
Depending on how long the application runs, starting and ending a
shadow batch job for each client may be more overhead than desired -
other things to consider are the working set (memory, open files) etc
of each shadow job. On the other hand, when serving many clients with
fewer server jobs, there may be throughput issues - eg input
transactions may back up on the server data queue if transaction rate
is high and the servers fall behind. Another issue to consider when
serving many clients with one server job is security - eg how does the
application currently ensure a client is authorized to database files?
Having server/daemon "equal opportunity" application jobs that can
serve many clients may also be attractive for supporting other user
interfaces such as PC or web clients. For example iSeries Access (aka
Client Access) has data queue APIs, and the Toolbox for Java also has
data queue access methods.
HTH
--
Karl Hanson
Frank Whittemore
Is there a book, a web site or any other form of documentation that
would further explain the approach that you recommend? Something with
examples of code would be very useful.
Please jot down some more clues right here in this thread for all to
see.
Thanks.
Philip Jamison
Thanks a lot for the reply. Exactly the information/help I needed to get
started.
Philip
emc...@yahoo.com
For 10 years, we have been submitting interactive jobs to batch once
the user keys the essential data. Why eat up interactive cycles when
you can submit the number crunching part of the program to batch.
So the users has to wait his/her turn in the job queue. If the job ran
on screen, the user would just go smoke while the job was running
anyway. Submit it to batch, let them go smoke, they come back their
job is finished.
On 30 Sep 2002 16:54:06 -0700, fra...@adelphia.net (Frank Whittemore)
wrote:
Chuck Ackerman
case, what should the users do while waiting for a green screen program to
finish? ;-)
chuck
Opinions expressed are not necessarily those of my employer.
<emc...@yahoo.com> wrote in message news:3da41da7...@news.vnet.net...
emc...@yahoo.com
Here at work its the same thing everyday 30 people standing in line
and praying to the great coffee god for the pot to not run dry before
their turn.
Its crazy here. Production employees are given a 10 minute break every
2 hours.
However if you smoke your allowed to go outside an additional 5
minutes every hour to smoke. But only if you smoke, non-smokers are
not granted the additional 5 minute hourly breaks.
On Wed, 09 Oct 2002 14:52:28 GMT, "Chuck Ackerman" <so...@nospam.com>
wrote:
kra
> (interactive) and 1-n server (batch) jobs. Each server job could have
> logic something like:
> LOOP:
Karl,
Dividing a 5250 program into two parts, and passing messages back and
forth via data queue, is generally NOT an effective way to trim
interactive workload.
The reason is that most interactive workload involves inquiry and
maintanance - moving data from a database to a screen and moving data
from the screen to the database. It takes about the same amount of
CPU time to move data to and from data queues as database files (I've
tested this).
This is NOT to say that some programs that interface with a
workstation shouldn't be divided into two programs, where one of them
would be submitted to batch, as mentioned by others on this forum.
One good example would be a screen that prompts for a record selection
criteria then scans a database to produce a report.
On the other hand, data queues are quite effective for applications
where a single instance of a server handles transactions for multiple
concurrent interactive users. It's not so much for saving CPU time,
but it can save a lot of memory. A single instance of a transaction
posting program may handle all requests for 100 concurrent transaction
entry clients, for example.
Nathan M. Andelin
www.relational-data.com
Karl Hanson
This really depends on the type of work done for each transaction - eg
between the time the user presses Enter (or F-key) and the time the
result/answer is shown. If the work borders on trivial - eg issue one
read-by-key to a database file - then I would tend to agree. If the
work is non-trival - eg issue an SQL query and update/delete multiple
rows of data - I would disagree, because the transaction involves (in
terms of cpu cycles) a LOT more than simply moving data between screen
and database. It's all about cpu instruction pathlength.
An existing application could be instrumented to determine the percent
of cpu cycles spent doing non-5250-related work for each transaction.
If that percent is high enough, I believe moving the "heavy lifting"
to batch (via data queues or similar transport method) WOULD be an
effective way to reduce interactive workload. In other words, if the
data queue message passing pathlength is significantly less than the
non-5250-related work pathlength, the application may be a good
candidate to split.
Somewhere in the continuum of transaction complexity - from simple
"read one record, display that record" to "generate bill-of-materials
report" - there is a threshold at which moving work to batch starts to
pay off (my opinion). Splitting out the "business logic" also
provides the opportunity to use multi-client server jobs, with
potential for client diversity etc.
--
Karl Hanson
Ed Jensen
exhausted, but there is ample unused "batch CPU" available,
why doesn't the iSeries make use of that unused "batch CPU"?
Is there a technical reason?
-Ed
Charles Wilt
interactive CPW rating of you box. But when you do, the CFINT task kicks
in and starts sucking down CPU cycles. At some point, all the CPU is
being used by CFINT and whatever you were able to press the interactive
workload too.
CFINT is IBM's way of charging extra for the interactive processing
power.
HTH,
Charles
In article <WRDp9.1280$YJ1.2...@ruti.visi.com>, eje...@visi.com
says...
Karl Hanson
Discussed/debated often in this newsgroup.. you could probably use
Google groups search with a search key such as "CFINT" to find past
threads. The most complete online documentation is likely in the
Performance Capabilities Reference - see Chapt 2 Server Model
Behavior.
http://www.ibm.com/servers/eserver/iseries/perfmgmt/resource.htm
--
Karl Hanson
René H. Hartman
Charles Wilt wrote:
>
> CFINT is IBM's way of charging extra for the interactive processing
> power.
>
...or discounting for batch processing. It's just how you want to look
at it.
--
Vriendelijke groeten / Kind regards
René H. Hartman
R.H. Hartman Automatiserings Consultancy
www.hac-maarssen.nl
-----= Posted via Newsfeed.Com, Uncensored Usenet News =-----
http://www.newsfeed.com - The #1 Newsgroup Service in the World!
-----== 100,000 Groups! - 19 Servers! - Unlimited Download! =-----
Edwin Davidson
the CPU. If it starts to overheat, then the chip will throttle down -
but if I cool it good enough, then all is well.
If I buy a RISC processor from IBM (such as for the AS/400), they
govern me down so that I cannot use 100% of the CPU.
I never understood this. The processor is sitting there, has the
capacity to do more, but IBM doesn't allow you to use it to it's full
potential.
I know they say it is because we only purchased so many CPW, but they
sold a processor capable of more. But I find this an odd way of
selling this type of hardware.
Is it really some sorta like RAIN process like in the PC world?
Tom Kobelt
avoidance. :) NO upright citizen of the AS/400 community would ever want
to be charged with tax avoidance by the big blue powers that be; however,
there is no law against minimization of tax (income tax, green screen tax
and otherwise)
In this case the green screen "tax" seems like the heavy reparation
payments that Germany had to pay to the allies. What war did we lose? I
guess AS400 green screen lost the technology war. Around 1993 there was
an article in Harvard Business Review on Technology suggesting that IBM
should license OS/400 to others! The idea was one method to expand the
use and acceptance of the then leading edge technology. Despite all the
marketing stuff IBM puts out it is still in the business of selling
boxes! The "tax" was devised by finance guys and accounts for the cost
for us to re write our applications to less efficient or effective
methodologies that run tax free on an AS/400.
Tax minimization strategies, which follow other basic design strategies,
isolate the user interface (screen) from the rules. Of late, I have the
screen call a separate function (program) for updating the data base etc.
You can then move that to make a request by batch submission or an entry
on a data queue. So an order entry screen program would run interactive
but the order update would be run in batch. Finally with the logic
separated from the screen you could use Net.Data or Web Fear (Sphere) to
call the function and then you would be at the 0% tax rate.
Finally, with the tax rate so high, IBM forces all of us to reconsider
applications completely outside of their tax jurisdiction (Port to another
platform). We have moved one application over (Contact Management) to
Cold Fusion and are now considering moving others.
Pieter
"Edwin Davidson" <edavi...@primeinc.com> wrote in message
news:11acaa5.02101...@posting.google.com...
René H. Hartman
There have been numerous threads on the subject in the past, so if you
Google a bit for CFINT you may find those.
Bottom line is: You pay for the performance of the box, not for the CPU
chip. You need high volume performance: get an expensive box. You need
little performance for your one-man shop: get an el-cheapo. Both boxes
may have the same chip, but that's not relevant.
In Intel terms: a 2.4GHz P4 is a lot more expensive than a 1.6GHz one.
By comparison: IBM only does the 2.4GHz chip, but if you only want to
run it at 1.6GHz, you pay the lower price. Then don't complain that the
system won't allow you to overclock it to 2.4GHz.
Ed Jensen
: Google a bit for CFINT you may find those.
: Bottom line is: You pay for the performance of the box, not for the CPU
: chip. You need high volume performance: get an expensive box. You need
: little performance for your one-man shop: get an el-cheapo. Both boxes
: may have the same chip, but that's not relevant.
:
: In Intel terms: a 2.4GHz P4 is a lot more expensive than a 1.6GHz one.
: By comparison: IBM only does the 2.4GHz chip, but if you only want to
: run it at 1.6GHz, you pay the lower price. Then don't complain that the
: system won't allow you to overclock it to 2.4GHz.
This is interesting, but it doesn't address my question.
My question assumes CPU power you *did* pay for is available
but unused. It just happens to be "batch CPU" power rather
than "interactive CPU" power.
I must admit, I'm still confused on the issue. And don't
fully understand why interactive jobs can't "dip into"
unused batch CPU, and why batch jobs can't "dip into"
unused interactive CPU. Once again: CPU you *did* pay
for. I'm not talking about selling 2.4 GHz CPUs as 1.6 GHz
CPUs.
-Ed
Edwin Davidson
is a good idea.
> In Intel terms: a 2.4GHz P4 is a lot more expensive than a 1.6GHz one.
> By comparison: IBM only does the 2.4GHz chip, but if you only want to
> run it at 1.6GHz, you pay the lower price. Then don't complain that the
> system won't allow you to overclock it to 2.4GHz.
No -- I am not paying the lower price. I am paying the full price
for the 2.4Ghz machine. It cost IBM $x to manufacture the 2.4Ghz
chip, and they are charging me way more than the normal cost of that
chip even when I am buying the CPW equal of the 1.66Ghz chip - because
it is IBM.
Last time I added memory to the box, the memory looked just like 32
PIN SIMMS - looked just like the memory I add to my PC's. But it
cost me over 10x more because it was IBM memory. The 512Mb worth of
SIMM's at the time were around $100 for the PC side, but cost $2500
for the IBM.
Now, I love the AS/400. But what makes the those chips cost $2500?
The same thing with the CPU. I know they are making a big deal in
marketing that you can purchase an AS/400, and if you need more CPW
you can simply call IBM, pay a few bucks, and they will let you run
faster for a while. There is something about this I just don't like.
---------------
That AS/400 I am talking about is powered off right now -- I just
wonder if I replaced those SIMMS with the same type of PC SIMMS if it
would IPL..?
Frank Whittemore
Unfortunately, the many links to URLs were lost when I copied my Inbox
E-mail over to this thread.
-------------------------------------------------------------------------------
FAST400 for your AS/400 and iSeries servers.
What is it?
FAST400 is a unique software utility to maximize interactive
performance of your AS/400 or iSeries server system.
What does it do?
FAST400 automatically tunes each interactive job on your system to
minimize the CPW resources used by the job. On server-class systems,
interactive workload is usually constrained to a value much less than
what the machine is capable of. FAST400 largely relieves those
constraints for your jobs to allow you to get maximum performance from
your system without the need for expensive extra Interactive Features.
With FAST400, you get the most out of your system whenever you need it
without having to install additional interactive features to cater for
peaks in your interactive workload.
Is it fail safe?
Certain PTFs may contain code designed to circumvent technological
measures in place to prevent tampering with FAST400's ability to
function correctly. In keeping with the spirit of IBM's eLiza project,
FAST400 is "self-healing" and "self-adjusting" and so, FAST400
automatically detects if such PTFs have been applied and adjusts
accordingly.
If you apply a new PTF, you may have to download the latest version of
FAST400. We try our best to be up-to-date, but there will always be a
delay of a few weeks. Should you come across a PTF that seems to
prevent FAST400 from working, please let us know.
Does it work with my system?
FAST400 only works on server RISC systems. Click here to see a
list of systems where FAST400 can help.
How does it work?
When your jobs perform workstation I/O, a bit is set in a field within
the description of each such job, marking the job as using interactive
CPW. Depending on your PTF level, FAST400 simply clears these bits
periodically, makes them invisible to the system, or tells the system
not to consider such jobs as contributing to the interactive workload.
How do I test FAST400 in my environment?
A free, FULL-function downloadable software product is available. The
demo code provided will limit the function of FAST400 from 09:00 AM to
11:00 AM each day according to your system clock. The Demo is
designed to allow your organization to see the benefits of FAST400.
After you purchase FAST400, you will receive another code so FAST400
will work 24/7 on your system.
Follow these steps to download your DEMO:
1. Download the FAST400 documentation below.
2. Download the FAST400 Save File below.
3. Use this sixteen digit code for your demo: 1234567890ABCDEF
Description
Downloadable
FAST400 Documentation:
fast400doc.html
7 KB
FAST400 Save File
fast400.sav
22 KB
How do I purchase FAST400 for my environment?
FAST400 is available for one-year subscriptions. For $1000 USD per
processor per year, you will receive a code for each system you
license. The code will be based on your renewal date, the number of
processors, and the serial number of the system where FAST400 resides.
Follow these steps to purchase FAST400:
1. Download and print the FAST400 Purchase Order/Invoice below.
2. Enter your system and company information and fax it to
1-618-355-1580.
4. Forward the Invoice to your Payables department.
5. After your Purchase Order has been received, your system codes will
be emailed to you within 1 business day (allowing for time zone
differences)
Description
Downloadable
FAST400 Purchase Order/Invoice
fast400poinv.html
35 KB
How much is it?
FAST400 is priced at an affordable US $1000 per year per processor.
Single processor systems are $1000 per year.
Multi-processors are $1000 per processor per year. (An 8-way is
$8000/year.)
NOTE: The purchased license code needs the correct number of
processors to function properly.
How can I check that it works?
You can use WRKSYSACT to check on the CPU usage of the various jobs
and internal tasks running on your system. The system task CFINT01
shows how much CPU time is spent processing interrupts and is
typically a few percent or lower. When you exceed the interactive
capacity of your system, the system generates extra timer interrupts
at a rate high enough to prevent you from using more CPU. At such a
time, the total CPU usage (the load) will go to 100% and the system
spends most of its time processing the extraneous timer interrupts
leaving your jobs little time left over. If you have an n-way machine,
there will be a CFINT task for each processor: CFINT01, CFINT02, etc.
When FAST400 is running you will see the CFINTnn tasks revert to a few
percent or lower, and your jobs be given more of the CPU.
If you do not have WRKSYSACT installed, we provide a work-a-like:
FASTACT. You can download a free demo of FASTACT by clicking on the
files below. The demo version only works between 09:00 and 11:00 AM.
Follow these steps to download your DEMO:
1. Download the FASTACT documentation below.
2. Download the FASTACT Save File below.
Description
Downloadable
FASTACT Documentation:
fastactdoc.html
5 KB
FASTACT Save File
fastact.sav
91 KB
You can purchase a perpetual license for FASTACT for $500 USD by
filling out the appropriate entry on the purchase order for FAST400.
(You can purchase only FASTACT if you don't need FAST400.) The
non-demo version of FASTACT will then be returned to you as
confirmation of your order.
Resellers
If you would like to resell FAST400 or any other of our products to
your customers, please email us here for details.
Questions? Technical support? Sales support?
If you have any questions, need technical support, have sales
questions, or just have comments (praise accepted), please email us:
here or phone (44) 7766 072 578
How to contact us
US Address and FAX number:
Fast400
PMB 122
3936 Highway 52N
Rochester, MN 55901
Fax# 1-618-355-1580
UK phone number:
Phone (44) 7766 072 578
Email: fas...@fast400.net
FAQs
Click here for a list of FAQs (Frequently Asked Questions).
The fine print
Products and company names mentioned on this website may be trademarks
of their respective companies and are the property of their respective
owners.
By downloading any software from this site, you agree not to
disassemble, reverse engineer, trace, debug, alter, or otherwise use
or subject the software to use in any manner other than for its
intended and stated purpose.
Walter
You compare the RAM of a iSeries with the cheap one in PCs.
If you want to compare it, then do it with the right thing: another server.
You won't get 512 MB RAM for a Compaq Proliant or so for 100$, if bought
from Compaq, which is the only way to keep the 4 hours repair contract.
I do not know the prices for IBM servers, because IBM PCs and PC servers
vanished from the list of allowed hardware here.
Walter
Charles Wilt
The batch CPW rating is 100% of the CPU. So for example, my 270 has a
CPW rating of 50/1070 (interactive/batch). This doesn't mean that the
box has 1120 total CPW, it actually only has 1070 of which a maximum of
50 can be used for interactive work. The remaining, 1020 to 1070
depending on how much interactive is used, is available for batch work.
Charles
In article <nBGq9.1475$YJ1.2...@ruti.visi.com>, eje...@visi.com
says...
MBBrutman
capacity. If you choose to use the box in a "batch" sort of way, you
can use all of the CPU cycles of the box. Batch cycles are
unrestricted, up to the physical capabilities of the box.
If you need "interactive" power, then you may be subject to a
limitation. The "interactive" portion of your work is not allowed to go
above a specific amount of CPW, which can be roughly translated to CPU time.
Yes, it does seem arbitrary. The hardware is capable of a certain
amount of work total, of which work that looks to be "interactive" is
limited to a fraction of the total. As for the reasoning behind the
pricing structure, I'll leave that to IBM marketing.
From a customer's view, the way to maximize your value is to use as
much batch CPW as you can, and minimize the interactive work. This can
be done by running batch-type jobs like web servers, WebSphere
(WebFear?), etc. Those batch jobs may still have interactive users -
those users are behind web browsers or they are client-server
applications, not 5250 users. And that makes the difference ...
Mike
Frank Whittemore
> Actually, batch processing can dip into unused interactive.
>
> The batch CPW rating is 100% of the CPU. So for example, my 270 has a
> CPW rating of 50/1070 (interactive/batch). This doesn't mean that the
> box has 1120 total CPW, it actually only has 1070 of which a maximum of
> 50 can be used for interactive work. The remaining, 1020 to 1070
> depending on how much interactive is used, is available for batch work.
------------------------------------------------
Charles -
You have exactly the same 270 that my site has!
What does your site do to maximize the use of Batch and minimize the
use of Interactive CPU? My site is mostly custom written RPG for
green-screens. Does that also describe your shop?
To all -
What I'm most interested in hearing more about is what mostly custom
written shops are doing to maximize the use of Batch and minimize the
use of Interactive CPU.
Charles Wilt
It isn't a concern to us. While we use green-screen RPG apps entirely,
I've only got about 20 users. Thus our interactive workload is pretty
small, so the 50 CPW is plenty. (and a nice improvement from our old
model 400 :-) The nature of our business simply requires more batch
processing than interactive.
We actually could have gotten away with a smaller box, in fact we were
looking at the 2431 with 30/465 CPW. But the powers that be decided to
bump up our requested machine to the next higher model. Since the 2432
was the box I original wanted, before going through the motions of
finding out the minimum we needed, I was happy to oblige.
Charles
In article <bcdcd347.02101...@posting.google.com>,
fra...@adelphia.net says...
Edwin Davidson
> written shops are doing to maximize the use of Batch and minimize the
> use of Interactive CPU.
According to WRKHDWRSC.
Processor feature code . . . : 22A7
Processor feature . . . . . : 2252
Interactive feature . . . . : 1516
Which, if I understand http://www-5.ibm.com/pl/as400/pdf/h270.pdf,
means my AS/400 (270) has a CPW of 950, all of which is batch. Which
is fine, because all of my screens to my programs are HTML, processed
by RPGILE CGI programs - which is considered batch to the AS/400.
Since I have no interactive CPW, you may understand my prior
questions/statements. Not that I need the interactive CPW on this
AS/400, however when the idler kick in once in a while when I am
programming it gets very slow.
Edwin Davidson
interactive users in batch.
Not an endorsement, just something I saw in an advert.
Chuck Ackerman
Here's where the iSeries and Intel machines differ.
I was at COMMON last week and heard Frank Soltis tell the gathered crowd in
a session that sometime in the future that ALL iSeries machines will be sold
with extra capacity so that you can use the extra capacity when you need it
and return it when you don't. It's like renting Capacity on Demand (CoD).
So, that means when you get your iSeries it will have extra RAM, extra disk
and extra CPUs for you to rent when you need them. You will rent them on a
"per day" basis, not by the hour or CPU cycle. That way when you need extra
umph at month-end you can get it for a day or two.
He said that this capability is actually in the architecture now, but the
hang-up is the back end accounting and billing chores that IBM needs to work
out.
So, here's my question: If you purchase a single CPU iSeries with 1 gig of
ram and 100 gig of hard disk and IBM delivers a 4 CPU iSeries with 4 gig of
RAM and 400 gig of hard disk, is that extra capacity yours to use FOR FREE
if you figure out some way to subvert IBM's billing system? If someone
comes out with "Really Fast400" would you use it? Would you feel that
Capacity on Demand is just something that should be ignored? By the way,
Soltis said that, yes, interactive CPW will be included in capacity on
demand, so you can get extra interactive umph when you need it. That's why
they're fighting Fast400 so vigorously.
Here's another analogy: A cell phone has the ability to let one talk 24
hours a day, 7 days a week. Yet one doesn't dare do that because the
purchased a plan that "free" minutes. If someone came out with a hack
called TalkFast400, that lets a person talk unlimited without paying the
phone company would it be ok to use it? Why not? Because it's illegal or
because it's simply wrong to violate the signed contract? If the answer is:
because it's illegal and it's ok to do it if you could without being
punished by the law then I must question that person's ethics and morals.
People with questionable morals should not be in a decision making position
in their company.
chuck
Opinions expressed are not necessarily those of my employer.
"Edwin Davidson" <edavi...@primeinc.com> wrote in message
news:11acaa5.02101...@posting.google.com...
Robert Comer
> comes out with "Really Fast400" would you use it?
Definitely not.
>Would you feel that
> Capacity on Demand is just something that should be ignored?
Definitely not.
> Here's another analogy: A cell phone has the ability to let one talk 24
> hours a day, 7 days a week. Yet one doesn't dare do that because the
> purchased a plan that "free" minutes. If someone came out with a hack
> called TalkFast400, that lets a person talk unlimited without paying the
> phone company would it be ok to use it?
Definitely not.
>Why not?
It is theft and theft is wrong.
>Because it's illegal or
> because it's simply wrong to violate the signed contract?
Both actually. Not to mention immoral.
>If the answer is:
> because it's illegal and it's ok to do it if you could without being
> punished by the law then I must question that person's ethics and morals.
> People with questionable morals should not be in a decision making
position
> in their company.
You don't have to worry about me or my company. <g>
- Bob Comer
"Chuck Ackerman" <so...@nospam.com> wrote in message
news:qxUs9.7559$884.44...@newssvr21.news.prodigy.com...
Chuck Ackerman
Actually, I'm not worried about most of the population. Most everyone wants
to do the right thing, it's the vocal minority that wants to break the rules
because they think the rules aren't for them. The majority of the
population just does the right thing and says nothing about it.
chuck
Opinions expressed are not necessarily those of my employer.
| >If the answer is:
kra
> hang-up is the back end accounting and billing chores that IBM needs to work
> out.
Now, if IBM's back-end accounting were only handled on an iSeries, the
billing would be in sync with the hardware - no delay ;-)
> It's like renting Capacity on Demand (CoD).
Which underscores IBM's philosophy of attaching a price to capacity
vs. attaching a markup over the cost to manufacture. Whether that's
good or bad, I'll leave to the economists!.
> is that extra capacity yours to use FOR FREE if you figure out some way to
> subvert IBM's billing system?
Now, that's a loaded question!
> If someone came out with a hack called TalkFast400, that lets a person
> talk unlimited without paying the phone company would it be ok to use it?
I question the validity of your analogy, since the phone company
obviously owns the network.
A better analogy would be if you had purchased your own office phone
network, then added a 3rd party product which overrode built-in
constraints imposed by the phone system manufacturer.
Nathan M. Andelin
www.relational-data.com
Chuck Ackerman
phone network, then added a 3rd party product which overrode built-in
constraints imposed by the phone system manufacturer."
Actually we have an ACD system (Automatic Call Distribution) installed for
our call centers. We pay for X number of licenses. They install a dongle
that limits us to X. I would never dream of buying a third party hack that
would allow us unlimited licenses.
So, here's a better analogy. You buy JDEdwards and pay for 30 licensees. If
you can hack the system to give you unlimited licenses, would you do it?
kc_pgmr
a glaring difference. In the case of both the JDE and the ACD analogies, you
are talking about a "license". In the case of the hardware, you are talking
about the purchase of an actual piece of hardware.
A license gives me the privilege of using the item. If I purchase a 400 from
IBM, is the hardware not mine?
BTW, I whole-heartedly agree in not "stealing" - just playing devil's
advocate ;)
"Chuck Ackerman" <so...@nospam.com> wrote in message
news:sZZs9.174$_B4.19...@newssvr14.news.prodigy.com...
Dan Hicks
> Please forgive me if I am missing the boat here but to me there seems to be
> a glaring difference. In the case of both the JDE and the ACD analogies, you
> are talking about a "license". In the case of the hardware, you are talking
> about the purchase of an actual piece of hardware.
>
> A license gives me the privilege of using the item. If I purchase a 400 from
> IBM, is the hardware not mine?
How does this differ from buying a copy of software? I buy the
physical package at Best Buy. Why can't I defeat any license
limitation in the software so I can use it "to its full extent".
--
Dan Hicks
The key to whatever success I enjoy today is: Don't ask. Do. --
Vikki Carr
René H. Hartman
kra wrote:
>> If someone came out with a hack called TalkFast400, that lets a
>> person talk unlimited without paying the phone company would it be
>> ok to use it?
>
> I question the validity of your analogy, since the phone company
> obviously owns the network.
Hi Nathan, not agreed. IBM obviously still owns the capacity you didn't
pay for.
Analogy's fine with me.
> A better analogy would be if you had purchased your own office phone
> network, then added a 3rd party product which overrode built-in
> constraints imposed by the phone system manufacturer.
>
>
> Nathan M. Andelin
> www.relational-data.com
--
Tom McGivern
The license comes as to what software group charges you're paying for. If
you purchase a 9406-270 Performance Group 05, but you buy a used processor
card on the street to add a second processor (now a P10).. yes all the
hardware is yours, but your only paying supportline, and software support on
a P05 machine.
If you purchase the box outright, and don't pay IBM for any further support,
then you ought to be able to do anything you want. Use it in an electronics
lab, get Linux to run natvie.. Wire it to hold a TB of Ram,
Whatever you want, as long as you're not reselling it, or paying IBM for
support on something it's not. (I'm hesitant to say you would be able to
run OS/400 on a larger machine, as the license you purchased was still for
the smaller P05 machine). Now on the otherhand, if you were to get Linux to
run native.. then again, you should be able to do what you want with the
hardware.
Tom
"kc_pgmr" <kco...@nodom.com> wrote in message
news:vESdnewsqOg...@comcast.com...
Dave Wallen
give everyone a higher power box throttled down?
Imagine people buying a lamborgini countach? (please forgive any spelling
errors, and also picking one company) , and all of them have the same engine
as the high-priced sportscar, but some of the cars are sold cheaper with a
different computer that slows the performance.
thus, Lamborgini is selling the full performance model for 220K, the
slightly slower model for 180K, and the "dog" for 120K.
All the same car, however..only difference is the computer running the
engine.
So, in order to make money, Lamborgini has to CHARGE someone for all that
metal. After all, the dog version still costs the company the same amount of
money to produce as the high performance version. This is not reflected in
the price, so the only answer is that they are overcharging everyone enough
to make money on the whole series. Not that they should not make money, but
they should not be able to charge the "dog" buyers enough to subsidize the
high performance buyers. (IMHO)
So, is the company ethically correct?
Garry Taylor
I think the issue depends on how the contract is written, if you
compare this to Fast400, I think it's morally, and maybe even legally
acceptable. I mean, if you own the AS/400, you have the right (IMHO)
to run any software you feel like, and if that includes Fast400, so be
it. If you have an iSeries with more processors, which still belong to
IBM, then they are not yours to use without permission, but if you
don't like that concept, you should'nt sign up to it.
Frankly I think if IBM got more realistic with pricing, then people
would not feel the need to go out and spend $1000 per processor per
year on Fast400 just to get the most out of hardware they already own.
I think most of us here don't especially want Open Sourced OS/400 or
commodity-style pricing, but only for IBM to price the iSeries
similarly to the pSeries or xSeries, with no small-print and no
obscene pricing for stuff which should be chucked in for free.
I realise I've gone OT here, but if IBM want people to abide by their
rules, they should make their rules a bit fairer and less inclined
towards IBM's bottom line.
Garry
kc_pgmr
right to use an intellectual idea. You buy an agreement that allows you to
use their software; you do not actually buy their software.
When you buy a piece of hardware, what comes in the box should be yours.
Granted, the difference would be in the installed OS and other LIC PGMs but
if IBM doesn't want me to use the hardware that I don't pay for, get it off
my property!
"Dan Hicks" <danh...@ieee.org> wrote in message
news:3DB497A0...@ieee.org...
kc_pgmr
"Tom McGivern" <t.mcg...@attbi.com> wrote in message
news:cs2t9.47701$md1.8253@sccrnsc03...
Bradley V. Stone
anything that is purchased faster than the "dog" is gravy.
The problem with this thread is that we're assuming IBM is thinking
logically. They're not. They have too many chiefs running the show,
so in the end instead of one good idea you have 100 ideas melded
together with synergy that create chaos. :)
They are simply taking advantage of their very very very loyal
customer base. Intel would be doing the same thing if they could.
You never really buy an AS/400... you rent it. :)
On Mon, 21 Oct 2002 22:31:22 -0500, "Dave Wallen" <t2e...@hotmail.com>
wrote:
Edwin Davidson
FAST/400 existed prior to this thread. Not that I would ever use it.
We have paid IBM many millions of dollars for the many AS/400's we
have had over the years.
In the past, I have always praised the IBM for the reliability and
security of the AS/400. However, currently the PC's are reaching the
ability to be as secure and reliable as the AS/400. I have Compaq
servers running Databases such as SQL, INTERBASE, and applications
such as Firewalls, security systems, etc. All of which have uptimes
that can go for hundreds of days. Usually they are rebooted to apply
patches, but W2k is getting much better at not requiring a reboot.
The AS/400 still requires an IPL on some PTF's, as well.
It seems that we are moving many of our applications from the AS/400
to the PC simply because they have become so much more reliable, and
the costs are so much less. Many of these systems had AS/400
competition which was way more costly. A properly secured NT/W2k box
is secure, regardless of all of these M$ patches that are coming out.
Most of them don't work if you don't have NBT enabled and don't use
IE/Outlook on the server.
I guess the biggest benefit to having the CoD feature (capacity on
demand) is that I don't have to bring the AS/400 down when I need
additional capacity. I no longer have to insert a processor card or
memory in the memory card, requiring an IPL. This does seem like an
area that AS/400 has an advantage over my PC servers. I just question
how they can do this and not have to charge their customers a lot more
for the addition of this extra capacity. I guess this just depends on
how many people they get to use CoD so they can spread the cost
around.
Time for my morning Caffine.
Edwin Davidson
Rodney Johnson
new function to OS/400, they drop OS/400 into maintenance only mode. That causes a drop in sales
because everyone believes the iSeries is nolonger strategic...IBM reacts by dropping OS/400
completely...cause and effect. Are all you of against IBM's changing in hardware packaging really
thinking about this?????
Garry Taylor wrote:
--
Rodney A Johnson
Technical Team Lead for AS/400 Spool
Dept GJC
IBM Rochester, Minnesota
The contents of this message express only the sender's opinion.
This message does not necessarily reflect the policy or views of
my employer, IBM. All responsibility for the statements
made in this Usenet posting resides solely and completely with the
sender.
Chuck Ackerman
The hardware may be yours, but the operating system certainly is NOT yours.
You license it. And when you buy a third party product to hack the
operating system you are breaking your license agreement.
chuck
Opinions expressed are not necessarily those of my employer.
"kc_pgmr" <kco...@nodom.com> wrote in message
news:vESdnewsqOg...@comcast.com...
charlieojr
wrote:
>I was at COMMON last week and heard Frank Soltis tell the gathered crowd in
>a session that sometime in the future that ALL iSeries machines will be sold
>with extra capacity so that you can use the extra capacity when you need it
>and return it when you don't. It's like renting Capacity on Demand (CoD).
What a dumbass idea. I really doubt IBM is going to eat the cost of
unused CoD, so this is just a way to double dip.
Dan Hicks
> What you purchased at Best Buy is not the physical package as much as the
> right to use an intellectual idea. You buy an agreement that allows you to
> use their software; you do not actually buy their software.
>
> When you buy a piece of hardware, what comes in the box should be yours.
> Granted, the difference would be in the installed OS and other LIC PGMs but
> if IBM doesn't want me to use the hardware that I don't pay for, get it off
> my property!
You're making an arbitrary distinction. If you buy a display card
for your PC and it comes with software, doesn't the license on that
software still apply?
--
Dan Hicks
How much more grievous are the consequences of anger than the causes
of it.
Dan Hicks
The Lamborgini is expensive in large part because a lot of hand
labor goes into each individual machine. Used to be that the cost
of a computer was in the silicon and gold and plastic and
fiberglass. But now the money is in the design -- the percent of
the total cost that goes into the design is often more than half the
hardware cost. This makes computer hardware more similar to
software than to automobiles.
In addition, there are economies of scale in processor production.
Especially since slow processors are usually just artificially
slowed fast processors (because the expensive design work from the
high-end processors is reused for the low-end to save money), it
often makes sense to make all processors physically the same and
control their speed externally. (Only one production line, only one
process to control, better yields, and hence lower per-unit cost.)
But if you charged the same price for all the processors, regardless
of the use, then the average Joe wouldn't be able to afford one,
and, as a result, the price would go up as well for the high-end (no
economy of scale). (Note that again this is a software-type
situation -- the per-unit cost is small compared to the fixed costs
that must be spread across all units.)
You can complain about the model, but without it you'd have to pay
MORE, not less, for every system.
--
Dan Hicks
Only cynics know the price of everything and the value of nothing.
--Oscar Wilde
Dan Hicks
> Frankly I think if IBM got more realistic with pricing, then
> people would not feel the need to go out and spend $1000 per
> processor per year on Fast400 just to get the most out of
> hardware they already own.
Hmmm. Then why doesn't someone write software to break the locks in
Fast400 so that you don't have to pay per-processor? Wouldn't that
be just as legitimate. How realistic is it to charge multiple times
for the SAME piece of software???
--
Dan Hicks
Sometimes it's a little better to travel than to arrive. --Robert
Pirsig
kc_pgmr
hardware without OS400?
"Chuck Ackerman" <so...@nospam.com> wrote in message
news:Gydt9.597$MO7.43...@newssvr13.news.prodigy.com...
Dave Wallen
Ok, bad example. I do have a habit of doing that.. ;-)
>
> In addition, there are economies of scale in processor production.
> Especially since slow processors are usually just artificially
> slowed fast processors (because the expensive design work from the
> high-end processors is reused for the low-end to save money), it
> often makes sense to make all processors physically the same and
> control their speed externally. (Only one production line, only one
> process to control, better yields, and hence lower per-unit cost.)
> But if you charged the same price for all the processors, regardless
> of the use, then the average Joe wouldn't be able to afford one,
> and, as a result, the price would go up as well for the high-end (no
> economy of scale). (Note that again this is a software-type
> situation -- the per-unit cost is small compared to the fixed costs
> that must be spread across all units.)
>
> You can complain about the model, but without it you'd have to pay
> MORE, not less, for every system.
I'm not advocating theft (and yes, that is what it would be), but I still
wonder... if all car engines were designed to have 8 cylinders, and you only
ran as many as you paid for (4,6,8), how many of us would be under the hood
with a schematic?
(Just curious about my fellow man ;-))
Tom Kobelt
on your boss, your spouse, and your clients.
Cheating is just taking a short cut. Ultimately we end up cheating on
ourselves.
The economics of building a processor (In a prior lives I worked in
Electronics Companies) compels us to gain economy of scale since the
design is a significant part of the cost. If IBM did not tier the
processors they would ultimately sell less processors and have less
units to write off their development costs. They have engaged in market
differentiation which utilities and other companies also use. I have
done the same thing myself. Two flavours of processor and the only
difference is what we burned on an EPROM. We saved on design and
production and ultimately could lower the cost of the product. IBM can
do this, because like a utility, they have some degree of control of
their market in the short term. In the long term there will be
alternatives.
I have no issue with the general architecture; however, if I can buy a
40 to 60 GB hard drive for a server or PC at the drug store for hundreds
of dollars. Then I can afford to buy two and mirror them against the
$2000 I pay for a 17 GB hard drive for the AS/400. A dual processor x
series with Linux looks very attractive. In the short term we are all
locked in by our prior investments. In the long term we can be making
investments to move to other platforms or we can decide to save the
porting cost and stay with the platform.
Ultimately these decisions have nothing to do with technology and every
thing to do with marketing and economics.
Do take a short cut on the vendor - NO - I will let the vendors slit
their own throat - YES.
Dan Hicks
> I have no issue with the general architecture; however, if I can buy a
> 40 to 60 GB hard drive for a server or PC at the drug store for hundreds
> of dollars. Then I can afford to buy two and mirror them against the
> $2000 I pay for a 17 GB hard drive for the AS/400. A dual processor x
> series with Linux looks very attractive. In the short term we are all
> locked in by our prior investments. In the long term we can be making
> investments to move to other platforms or we can decide to save the
> porting cost and stay with the platform.
I agree that the hard drive situation is silly. I don't think that
IBM is making that much off the drives (all that Lamborgini hand
crafting), and it wouldn't be too hard with RAID technologies to
design a controller that could provide virtual 520 byte sectors on
standard drugstore drives.
--
Dan Hicks
Nothing can bring you peace but yourself. --Ralph Waldo Emerson
Garry Taylor
I would hope that a drop in price would encourage more sales, not just
reduce income on existing sales. But the hardware is a lesser issue
than software (although iSeries pricing in line with pSeries would be
great), but the software pricng is far worse, if IBM wants to
encourage more development on the platform, it should make development
tools cheaper, or try to get gcc working on the AS/400. The fact the
Java is free on the iSeries is great, but not everyone likes it or
finds it suitable for their work, free RPG and C on the other hand
might get more people porting UNIX apps, and writing from scratch for
the platform. I don't use the AS/400 at work, and am merely a
developer who likes the look of the AS/400, I can get on one eBay for
about $1000, and that's great but I'd have to pay $1000 again just to
get the programming tools I need to do anything with it.
I'm not asking for IBM to sell stuff at cost, or to drastically cut
prices, all I want is for them to give comparable value to their own
UNIX kit, and also comparable to Sun, HP and the rest.
Garry
Rodney Johnson
expensive middle and upper management...) perhaps it would be possible to reach prices more in tune with
what you would like to see. But IBM is a big organization with a lot of overhead. The money has to come
from somewhere to support their salaries and benefits. Unfortunately that ends up rolling down to the
consumer.
The costs of an integrated OS also does not help. None of IBMs customers realize just how expensive it is
to continue to enhance and add to an integrated OS.
As you might notice I am not a fan of big organizations, but am forced to live with it if I want to do
OS type development...and I sure don't want to work for M$. I don't agree with many decisions that are
made especially when I see good developers get a pink slip because the bottom line wasn't good enough at
the end of a quarter. And its not like the work goes away...it is still there and still has to be
done...just with fewer developers.
Chuck Ackerman
Here's a real example. Many people lease a car that requires them to pay
mileage fees for any usage over that. However, I don't expect many people
rush to rollback their odometer before turning in their car. They just pay
the penalty fee and move on. If there were excess cylinders, the same
would be true.
chuck
Opinions expressed are not necessarily those of my employer.
"Dave Wallen" <t2e...@hotmail.com> wrote in message
news:3db601ac$0$1425$272e...@news.execpc.com...
Chuck Ackerman
cheat
on your boss, your spouse, and your clients. Cheating is just taking a short
cut. Ultimately we end up cheating on
ourselves."
Speak for yourself. It is NOT okay to cheat on your spouse, boss or
employer. I hope you were speaking toungue-in-cheek here.
Tom also said:
! In the short term we are all
| locked in by our prior investments. In the long term we can be making
| investments to move to other platforms or we can decide to save the
| porting cost and stay with the platform.
I've been in top I.T. management in shops that have both AS/400s and
Linux/Solaris equipment. And, I can tell you, without any hesitation, that
the AS/400 shop is MUCH CHEAPER to manage. You've simply got to understand
that while Linux itself is cheap, er free, good hardware isn't. And, more
importantly, Linux shops require more support personnel than an AS/400.
From programming to operations, to DBAs the Linux environment costs more.
I'm not talking about a small Linux vs. small AS/400 shop, I'm talking about
a $300 million plus company with more than 2,000 employees and dozens of
Linux boxes that try to do what a single AS/400 does. The Linux shop will
have triple the personnel of the AS/400 shop. In fact, the Oracle DBA staff
will probably be as large as the entire AS/400 staff.
chuck, speaking from experience.
Opinions expressed are not necessarily those of my employer.
|
Tom Kobelt
It may sound like I am advocating cheating - when of course I gave the
consequence - I thought I would level the playing field by extending the logic.
If it is okay to cheat a vendor then it would follow on that you could cheat
others. An issue with employee theft is that it starts with a small
rationalization and then continues. Ultimately such behaviour results on
cheating on yourself. I limit my comments here since the context of this
newsgroup is technology and not philosophy.
Chuck Ackerman
technology and not philosophy."
Of course, the simple choice of a technology is a philosophy. Or, as some
would say, religion.
chuck
Garry Taylor
> Garry Taylor wrote:
>
> > Frankly I think if IBM got more realistic with pricing, then
> > people would not feel the need to go out and spend $1000 per
> > processor per year on Fast400 just to get the most out of
> > hardware they already own.
>
> Hmmm. Then why doesn't someone write software to break the locks in
> Fast400 so that you don't have to pay per-processor? Wouldn't that
> be just as legitimate. How realistic is it to charge multiple times
> for the SAME piece of software???
That's a good question, I think on machines like the iSeries,
per-processor payments are the fairest way to price, as it means that
software developers can let buyers with smaller boxes have the
software at a lower price, rather than simply find a mid-way price
which is too much for the smaller shops. I can't say I like
per-processor pricing, and I definitley don't like paying annually for
software, but I can't see a good way round it.
I guess if you signed a contract with IBM saying that you would'nt use
Fast400, then the contract is binding, but if you have not made any
kind of formal agreement with them, then I think it's legally and
morally acceptable to use any software you like.
I accept the argument from Rodney Johnson that IBM must pay for
development, and that iSeries development is expensive, and these
costs result in iSeries machines costing more than pSeries, however I
think a bottom end pSeries rackmount is about $2500, and a bottom end
iSeries is a little less than $20,000, and of course, a C compiler on
the iSeries is going to add to that. I have not checked out the specs
of either machine, so maybe the iSeries has loads of features (like
DB2) that make it worth the money, however I think it does illustrate
that those of us who think iSeries is bit expensive may have a
legitimate beef.
I don't think it's OK to pirate software, nor to cheat vendors, but I
don't think IBM should be extorting a great deal of money from it's
loyal customers for things like 'interactive features' for what costs
IBM pretty much nothing.
Just my two cents
Garry
Chuck Ackerman
I haven't read my contract in a long time. But I can guarantee you there's
a clause in the contract allowing them to alter it in the future. I suspect
they alter it in the PTF cover letters or in the memo to users in an
upgrade.
chuck
Opinions expressed are not necessarily those of my employer.
"Garry Taylor" <gta...@lowebroadway.com> wrote in message
news:f0fd5987.02102...@posting.google.com...
Frank Whittemore
exchange with IBM using the Internet IBM SupportLine function:
Problem: What, if any, programming techniques does IBM recommend to a
mostly custom coded RPG legacy AS/400 green-screen shop in
order to move more work from QINTER to QBATCH over time to
avoid the governor?
.
Action Taken: Hi,
Thank you for using IBM Electronic Support. In general we do not make
recommendations for programming techniques in Support Line.
Determining
the best programming technique would require a detailed analysis of
the
current environment and all programs running on the system. Such a
detailed analysis of the system environment would best be handled by
highly skilled technicians through our Consult Line offering or by our
Global Services division.
.
If you have a specific question on a specific RPG function we will be
glad to assist you further.
.
charl...@yahoo.com
>
> The Lamborgini is expensive in large part because a lot of hand
> labor goes into each individual machine. Used to be that the cost
> of a computer was in the silicon and gold and plastic and
> fiberglass. But now the money is in the design -- the percent of
> the total cost that goes into the design is often more than half the
> hardware cost. This makes computer hardware more similar to
> software than to automobiles.
>
> In addition, there are economies of scale in processor production.
> Especially since slow processors are usually just artificially
> slowed fast processors (because the expensive design work from the
> high-end processors is reused for the low-end to save money), it
> often makes sense to make all processors physically the same and
> control their speed externally. (Only one production line, only one
> process to control, better yields, and hence lower per-unit cost.)
> But if you charged the same price for all the processors, regardless
> of the use, then the average Joe wouldn't be able to afford one,
> and, as a result, the price would go up as well for the high-end (no
> economy of scale). (Note that again this is a software-type
> situation -- the per-unit cost is small compared to the fixed costs
> that must be spread across all units.)
>
> You can complain about the model, but without it you'd have to pay
> MORE, not less, for every system.
Artificially slowed fast processors is a little misleading. IBM's CoD
crap aside, usually a line of processors come off the same production
line. They are then bench tested with the speed ramped up over and
over until the processor begins to malfunction. At whatever speed the
malfunctions began to occur, the processor gets dropped into the bin
of processors rated at the next lower increment. Thus, your pentium 4
2.8Ghz CPU may have been right next to your neighbors 1.4Ghz on the
wafer.
The difference in price between the two is not based on cost but
demand.
Dan Hicks
>
> Artificially slowed fast processors is a little misleading. IBM's CoD
> crap aside, usually a line of processors come off the same production
> line. They are then bench tested with the speed ramped up over and
> over until the processor begins to malfunction. At whatever speed the
> malfunctions began to occur, the processor gets dropped into the bin
> of processors rated at the next lower increment. Thus, your pentium 4
> 2.8Ghz CPU may have been right next to your neighbors 1.4Ghz on the
> wafer.
They'd like you to believe that, and there is a little truth in
early production. But once the process is "tuned" there will be few
"slow" units and the speed limitation will be totally artificial.
--
Dan Hicks
I have abandoned the follies of youth for those of old age. -Depew