UBE's on the 400 & IBM O/S 400 support & DASD PTF's

Michael L.

Well Known Member
UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

It has been my experience that UBE's run well on the AS/400 and/or iseries
400 but only if configured properly. In other words, think about creating
an additional batch subsystem (try not to start with the letter "Q", this is
reserved for IBM objects) for the UBE's and "DEDICATE" an appropriate amount
of memory to that subsystem.

Try to ensure your total % DASD utilized does not exceed 70% (wrksyssts, top
right hand corner). Keep your O/S current, Keep your O/S current and Keep
your O/S current. Review and implement (carefully) the performance tuning
recommendations on the Knowledge Garden. I believe there is also a white
paper(s).

If you have the LPP performance tools (5769PT1) execute and review the
performance reports against a job, user, subsystem, overall system etc.....
I usually "fire up" this tool when the CPU is very high, but you can execute
anytime. Maybe you want to execute when a slow performing UBE is executing.
You could then print (or review online) exactly what that UBE is doing.
EXCELLENT tool if used properly and well worth the exercise.

An example
========

1) go perform
2) select option # 2
3) select option # 1
4) select option # 2
5) Try to use QPFRDATA for the library
6) 1 and 1 would be good
7) 15 & 1 would be good
8) your call

Remember, O/S 400 "out of the box" is not optimally configured for superior
OneWorld performance.

I have noticed several AS/400 configuration that are still on O/S 400 V4R3
and V4R4. See below, get current, it's important.
 
Re: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

Are you saying UBE's will run just as fast or faster on an AS/400 as they do on NT? If so, this is the first time I've heard that. Most, if not all the opinions on the list disagree with this. Maybe there's hope yet!

Ryan Allen

OW Xe B7333 SP15 / World A7.3 cum 12
ES - AS/400 V4R4
DS - Win2000
Fat - Win200
TSE/Citrix - Win2000
 
Re: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

FYI, I've never seen a good apples to apples comparison, it's tough to set
up a good test to compare a 400 and a comparably powered NT box, similar
processor power, same network configuration, same data, same
indexes/logicals, etc, so I'm not aware of any real benchmark type testing
that has been done.

The only test I've seen, which was pretty informal and wasn't the most
scientific, but it was at least more like comparing tangerines and oranges,
than apples and oranges...the 400 was faster. Maybe it was just tuned
better!

Rob Jump
Sizing Specialist
IBM/J.D. Edwards International Competency Center
303-334-1054
[email protected]


Spike <[email protected]>@jdelist.com on 07/19/2001 03:04:12 PM
 
Re: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

I know I can't compare an AS/400 to a comparable NT box, but I would think the model 730 I'm using here should be able to process jobs as fast as my local PC or at least close to that. Again, my AS/400 has plenty of resources available. In World, jobs process well, if not great on my AS/400, and it appears the OneWorld jobs don't. I just can't find anything else to tune on the AS/400 to improve this situation. I was hoping to find out if other OneWorld customers have found the same results as I have found.

Ryan Allen

OW Xe B7333 SP15 / World A7.3 cum 12
ES - AS/400 V4R4
DS - Win2000
Fat - Win200
TSE/Citrix - Win2000
 
re[2]: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

Rob,

I'm sure the list would appreciate any of the sizing/tuning/optimization helps/items that ATS and IBM have done to make UBE's run faster. I know from a user standpoint, UBE's almost always run faster on a wintel (desktop or server) machine than the 400.

FYI, I've never seen a good apples to apples comparison, it's tough to set
up a good test to compare a 400 and a comparably powered NT box, similar
processor power, same network configuration, same data, same
indexes/logicals, etc, so I'm not aware of any real benchmark type testing
that has been done.

The only test I've seen, which was pretty informal and wasn't the most
scientific, but it was at least more like comparing tangerines and oranges,
than apples and oranges...the 400 was faster. Maybe it was just tuned
better!

Rob Jump
Sizing Specialist
IBM/J.D. Edwards International Competency Center
303-334-1054
[email protected]


Spike <[email protected]>@jdelist.com on 07/19/2001 03:04:12 PM




--------------------------
 
Re: re[2]: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

I would love to be able to provide any insight. We usually try to put any
tips or info we find on either our website,

http://www.developer.ibm.com/erp/jdedwards/index.html,

or on the IBM-JDE support website,

http://www.ibm.com/servers/eserver/iseries/service/bms/jde-support.htm,

or in redbooks or on the JDE Knowledge Garden.

Unfortunately, I don't have much at all regarding UBEs. A lot of our focus
in on user performance and general system tuning. There is info on
webserver tuning and more user related info, but we have not done many
real, structured tests with UBEs. We have done some to look at performance
on a single processor, and the impacts on the interactive workload, but we
were looking at relative performance, and not really at the performance of
specific UBEs or trying to tune the UBEs. We also do some for release
testing, but again, they are focused more on comparing releases and
relative performance than trying to get the UBEs to fly.

From what I've picked up in passing, as Michael Lockwood pointed out, a lot
of it is 400 performance tuning; separating out different workloads into
separate pools, checking performance reports to look for bottlenecks, like
disk, memory, communications, or looking at indexes on the database side.
Also, as Michael points out, the PTF and software levels can be pretty
important.

It also seems to be very variable, depending on WHICH UBE is being talked
about. Some are slower on certain platforms, some faster. I'd love to
have information on which perform better where, but I haven't heard of
anything like that being put together.

Sorry we don't have more info from our area. JDE might have more on the
Knowledge Garden, but I'm not sure. We have more UBE testing planned, and
hopefully we can dig deeper and get better information, but right now we
don't have much at all.

Rob

Rob Jump
Sizing Specialist
IBM/J.D. Edwards International Competency Center
303-334-1054
[email protected]


Mike_Dupaix <[email protected]>@jdelist.com on 07/20/2001 10:44:57 AM
 
RE: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

I have the same complaint on our 730 as well. We have tuned about as much
as I can think to ask about and have had a number of IBM and JDE system
types in to help as well. What really helped us was using logging that
detailed what SQL would recommend for keys. That process took a few days
but was the biggest win we've had. I am still looking for another magic
bullet but I'm not sure one exists. We are not however running coexistance.
The biggest issue we have right now is CSMS. Our phone sales folks are
experiencing 30-45 second response times between screens which for phone
sales is unforgivable. All hints gladly accepted.

Darrell Allison
 
Re: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

I know about the IBM Redbook and oti-00-0107, What else has good infomation about tuning the AS/400?

Frank

OneWorld Xe + XU1 SP 15_010, AS/400 V4R4M0
Windows 2000 SP1, SQL2000 SP1,
Citrix Metaframe 1.8A SP1
 
RE: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

Darrell,
One of our clients had this same problem. We created a join logical
between the CSMS and Address Book files and our response on the screen
went from several minutes to under 6 seconds. This was on a co-exist
AS/400 that was very slow (400+ CPW) with about 200+ calls a day.

Good luck

Russ Stenquist
The MASYC Group, Inc.
WWW.MASYC.COM
(714) 479-0900
 
Re: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

We have a consultant who was converting payroll info over from our old
system to JDE using a z-file. We have an S20. He calculated based on the
first run that the second run would take about 100 hrs. After 4 days, it
was only about 1/8 done. He killed that project and ran one of his own that
finished in about 2 hours. We are thinking that interactive runs faster
then batch w/ our box. On the old (legacy) system it was the other way
around. I'm wondering if it isn't just a tuning or maybe an XE problem?


----- Original Message -----
From: "AllisonD" <[email protected]>
To: <[email protected]>
Sent: Friday, July 20, 2001 4:00 PM
Subject: RE: UBE's on the 400 & IBM O/S 400 support & DASD PTF's


coexistance.
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number=
17056



Jeremey Garcia
Xe SP14 ES - AS/400 CO - AS/400 Deploy - NT Citrix
 
Re: re[2]: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

Perhaps you could give some of the relevant details of your opinion.

I am quite sure my 8-way 64-bit AS/400 with 8 GB RAM can run a UBE over my
27.8 GB production database several times faster than your Compaq Dell 8-way
32-bit with 8 GB with 15 GB production database.

The devil is in the details pal. Talk apples to apples or noone will take
you seriously.


----- Original Message -----
From: "Mike_Dupaix" <[email protected]>
To: <[email protected]>
Sent: Friday, July 20, 2001 11:44 AM
Subject: re[2]: UBE's on the 400 & IBM O/S 400 support & DASD PTF's


helps/items that ATS and IBM have done to make UBE's run faster. I know
from a user standpoint, UBE's almost always run faster on a wintel (desktop
or server) machine than the 400.
oranges,
http://198.144.193.139/cgi-bin/wwwthreads/showflat.pl?Cat=0&Board=OW&Number=
17015



JD Nowell
OW: B7332
ES: AS400 V4R4 CO: DB2/400 SP: 11.2
Users: 250 TSE Users: 100
 
RE: re[2]: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

Guys :

I wouldn't bet my life on what machine would win... AS/400 or NT?
OneWorld performance on AS/400 is disappointing when compared to NT or Unix,
even on machines with significantly
poorer technical specs. I don't know exactly why it happens, but it
definitely does.
Of course, I fully agree that AS/400 stability and quality is far higher
than NT; but this time we're not talking about that, just
about "speed" : How much time does it take to process this specifical UBE
with this data? Try the following :
Add a Pentium III 500 Mhz + 512 Mb RAM Windows 2000 Server (a large PC) as
an UBE Server to your installation, don't
install any DBMS there (except for the SVM7333 tables). Just let it access
Business Data on the AS/400 via Client Access.
Run a given UBE on the AS/400 accesing its own local data (DB2/400). Write
down the elapsed time to complete that.
If you want to be absolutely sure that no cache effect will interfere and
favor NT, then IPL your AS/400 and your UBE Server.
Submit the same UBE with the same data to the NT Box, and write down the
elapsed time too.
You would be surprised by the results...

Sebastian





B7321 to Xe, NT/W2K/SQL
JAS, Interoperability
MCDBA,MCP+I,MCSE,Citrix Admin
[email protected]
Grupo ASSA - Application Software SA
 
Re: RE: UBE\'s on the 400 & IBM O/S 400 support & DASD PTF\'s

How many processors on that box? We started off with 2 processors. We learned that IBM recommends that have have at least 1 more processor than the # of UBEs that you have running. Oh.... So we get to run 1 at a time.....

So we upgraded to 4 processors. Made a difference. We still run more UBEs then we have processors at a time, especially at night when there aren't any interactive users. During the day we typically don't have more than 3 running except for short bursts.

But when you get more UBEs running than processors, utilization hits 100%..

We use the original 2 processor box for development and I can tell you, it's SLOW with only 1 or 2 of us on it...

Hope this helps...


Dave Schlieder

B7332 SP 15 AS/400 NT
 
Back
Top