cancel
Showing results for 
Search instead for 
Did you mean: 

JMS to RFC without ccBPM ????

Former Member
0 Kudos

Hi all,

We have a scenario that uses the JMS Request/Reply towards the SAP R/3 4.6C (RFC), so we use the BPM to connect the JMS request (with correlation ID) to the sync RFC and then to the JMS Reply (pass the correlation ID), it works but <u><b>we have very bad performance</b></u>, does someone knows other ways to connect the JMS to RFC ?????

or maybe you have any idea regarding the performance ????

Regards,

Yaki

Accepted Solutions (0)

Answers (4)

Answers (4)

stefan_grube
Active Contributor
0 Kudos

Hi Yaki,

first you should check, if the sender of the scenario could be enhanced to be able to send more data at once. If the sender would not send single items, but a bundle of 1000 items at once, the performance would speed up.

The second approach is getting rid of the BPM. Create two RFC. The first RFC is call from the XI async. The interface should have the same fields as the JMS sender. The RFC inside calls the second RFC async. The interface of this RFC should have all parameter as the JMS receiver.

Regards

Stefan

Former Member
0 Kudos

I have achieved this without ccBPM

bhavesh_kantilal
Active Contributor
0 Kudos

Peter,

Can you elaborate on how you did that?

Regards,

Bhavesh

Former Member
0 Kudos

Hello Peter,

Can you share us with your solution ?

Regards,

Yaki

moorthy
Active Contributor
0 Kudos

Hi,

Performance depends on many parameters-

- Based on the design - e.g Transformation step is inside the BPM. If it is not required inside the BPM, try to avoid that

-Hardware Sizing-Memory Settings/VM memory settings. Sometimes patches/notes needs to be upgraded.

-

Refer this Performance Tuning guide-

https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/70ada5ef-0201-0010-1f8b-c93...

Hope this helps,

Regards,

Moorthy

MichalKrawczyk
Active Contributor
0 Kudos

Hi,

>>>> it works but we have very bad performance,

there are so manz wazt that influence this that it

it hard to name all of them

you need to check which of the steps takes too much time send receive transform

and then decide what to do - new patch, change memory seetings, more memory, other parameters etc..

there realy is much you can do

btw

of course it's very good to use a stress test tool

to stimulate a much bigger number of messages then you have now...

the performance issues are those where the fun begins

Regards,

michal