Slow Response since going to AS/400 570s

We have recently went to a 570 and we are having issues with the system taking a long time bringing back certain JDE screens, Order Entry, Stock Availability, etc.

Our response time on the batch jobs have gotten better but the interactive jobs appear to be taking longer. Our package builds are taking a shorter time to build on the server, as well.

We are looking for suggestions on what things we may need to change. Please help. Thank you

We are on 8.10 - ToolsRelease 8.93 S_1 - 570 V5R3 -
Client Access SI20055 - Citrix MetaFrame XP FR3, SP4

FatManTLA,

A few questions:

1) How many CPU's were on the box you came to and how many on your new 570?

2) What is/was the CPU utilization on your new/old boxes.

3) What is your setting for system value QQRYDEGREE?

4) Are we talking Fat Clients, Citrix Clients, or Web clients, and if HTML where are you running WAS?

We had extensive talks with IBM before specing our 570 which will be coming on line this fall.

You need to back off of iSeries Access SI20055 and back onto SI10817.
There are issues on this PTF and you'll see it as soon as you get back
to 10817. Don't know all the details but search for the earlier thread
on this and you'll get some background.

I do not have all the answers to your questions at this time. I will have all of them answered hopefully by the end of today. I know we are running 2 CPUs on our new box and we were running 3 on the old. All the slow issues appear to be coming from the Citrix clients. We do not use Web clients.

What should the setting for QQRYDEGREE be?

I will answer the rest of the questions later.

Thank you

As always the answer is 'it depends' In our case we run a large number (> 7000) ODBC jobs, and normally have several hundred SQL's actually running at one time. If you have it set to *OPTIMIZE many of these SQL's will spawn multiple tasks under the covers. If you get too many tasks for the number of CPU's you are using you will start to get CPU queueing. In your case going from 3 to 2 probably shouldn't cause any issues as the 570's CPU's are so much faster and can multitask.

One other question are you using the QAQQINI? If so make sure that IGNORE_DERIVED_INDEXES is set to *YES. This will allow you to send more to the new query engine.

I assume you have your BSFN's set to *LOCAL for your W environments?

Tom

Hi Tom.

Is that QAQQINI setting good for V5R3 as well as V5R2? I search on the IBM
site and it comes up with V5R2 references but no V5R3 ones. I would as sume
so but would like confirmation.

Thanks.





Tom_Davidson
<Tom.Davidson@joh
nsondiversey.com> To
Sent by: [email protected]
jdelist-bounces@j cc
delist.com
Subj ect
Re: Slow Response since going to
05/02/2006 10:33 AS/400 570s
AM


Please respond to
JD Edwards=AE
EnterpriseOne
<jdelist@jdelist.
com>






As always the answer is 'it depends' (Embedded image moved to file:
pic27412.gif)In our case we run a large number (> 7000) ODBC jobs, and
normally have several hundred SQL's actually running at one time. If yo u
have it set to *OPTIMIZE many of these SQL's will spawn multiple tasks
under the covers. If you get too many tasks for the number of CPU's you are
using you will start to get CPU queueing. In your case going from 3 to 2
probably shouldn't cause any issues as the 570's CPU's are so much fast er
and can multitask.

One other question are you using the QAQQINI? If so make sure that
IGNORE_DERIVED_INDEXES is set to *YES. This will allow you to send more to
the new query engine.

I assume you have your BSFN's set to *LOCAL for your W environments?

Tom


OW 7334 SP 23K, AS/400, WIN2K, TSE 2000, Metaframe XP FR 1, CO SQL 7.0, Two
Instances English/Western European & Japanese

It SHOULD be a no-op. But a friend, who should know, mentioned that if it was him he would set it.

When we goto 5.3 we will have it set, unless IBM publishes that the entry should be deleted.

Tom

I only ask because my customer is at V5R3 and the setting is NOT there by
default. We would have to add it to the QAQQINI table if we are to use it.

Thanks!






Tom_Davidson
<Tom.Davidson@joh
nsondiversey.com> To
Sent by: [email protected]
jdelist-bounces@j cc
delist.com
Subject
Re: Slow Response since going to
05/02/2006 11:14 AS/400 570s
AM


Please respond to
JD Edwards®
EnterpriseOne
<jdelist@jdelist.
com>






It SHOULD be a no-op. But a friend, who should know, mentioned that if it
was him he would set it.

When we goto 5.3 we will have it set, unless IBM publishes that the entry
should be deleted.

Tom


OW 7334 SP 23K, AS/400, WIN2K, TSE 2000, Metaframe XP FR 1, CO SQL 7.0, Two
Instances English/Western European & Japanese

We have our QQRYDEGREE value set at *OPTIMIZE

Hi, we had similar problems here after upgrading to 8.11sp1 from XE. We have a model 550 with 2 processors active. I was able to improve interactive performance by prestarting more QZDASOINIT jobs. We usually max out at about 175 users on at a time. I set our system to prestart 150 of the QZDASOINIT jobs and performance was much improved. It looked to me like in many cases, ODBC jobs had to be started to service the request. By prestarting we save the time it takes to 'startup' the job. Hope this helps. Good luck with your performance tuning.

Thank you for all the responses.

The processor multi-tasking hardware setting on our new 570 was turned off. We turned on the setting and response time got better but we are still experiencing some slow response times but not as often as we have in the past.

Hey Tom,

Do you think he would benefit from our database/ASSESSMENT??

Jen

We have done a few things to improve our response time. We have prestarted more QZDASOINIT and we made the processors use multi-tasking. The system seems faster but there are occasions when we get a slow response issue. I would like to thank everyone who responded with suggestions.