Re: Cobol programming issues

If it is OPM COBOL (sigh) then activation groups don't come into it.

You didn't say how the programs return - preserving the run unit would help in performance terms - but it is not an answer to the problem.

As to the other information - I suspect that you are possibly dealing with a queuing scenario if the OPM COBOL is running in a single job. However, while the information you provided might trigger some thoughts for the performance gurus, this is not really the right list to use. The COBOL group is a tiny tiny subset of the IBM i community here - I suggest that you repost this question on Midrange-L because that is where the performance folks hang out - here on the COBOL list very few of them would see it.


On 2014-05-19, at 12:32 PM, Imad Moukaddam <Imad-Moukaddam@xxxxxxxxx> wrote:

It is OPM COBOL.

no there is no record locking issue here, and I have checked the activation group. It is OK.

But one thing, I don't know if this could help:

as we have opened a pool of socket connections, it means that one JOB on the iSeries could be used over and over again, by different users. We have noticed the performance issue starting after 70-80th connection, as if the resources allocated to this job are over the limit, while the CPU is OK, RAM is OK, but as if the resources of this job became limited, causing a performance issue for all users connected to this job. We are sure that a job is used only after the previous user has finished his work, no intersection between jobs.

It is only a possibility, we are not certain of this, but for iSeries experts, maybe it would mean something...


________________________________
De : Imad Moukaddam
Date d'envoi : lundi 19 mai 2014 17:57
À : java400-l@xxxxxxxxxxxx
Objet : Cobol programming issues

Hi

Well this is a difficult one to explain :
We have developed a web application (JAVA) that launches COBOL programs through SOCKET programs. Everything is working just fine.
When we started to increase number of users through a performance tool, we have noticed a strange behavior on the COBOL programs, as they were running very slow (for some transactions, average time was in the range of 10-15 sec.). Upon drilling down, we found that a simple READ statement (COBOL side) is consuming about 110 ms, which is huge comparing to the normal status (average time was less than 1 ms normally)!

We have checked the following:
CPU is OK
Memory is ok
No locks (we are reading PF in OPEN OUTPUT mode)
The hard disk is a data center, known to support heavy usage

Any ideas?
Thanks

Imad


[http://www.sab2i.com/fichiers/20130204105607_ecologie.png] Pensez à l'environnement, n'imprimez ce message que si nécessaire ! Please, consider the environment before printing this email.
--
This is the Java Programming on and around the IBM i (JAVA400-L) mailing list
To post a message email: JAVA400-L@xxxxxxxxxxxx
To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/java400-l
or email: JAVA400-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives
at http://archive.midrange.com/java400-l.

Jon Paris

www.partner400.com
www.SystemiDeveloper.com