cancel
Showing results for 
Search instead for 
Did you mean: 

IDoc to IDoc Scenario with High Volumes

Former Member
0 Kudos

We have requirement to load around 2.4 million idocs using XI from source SAP to target SAP system in ~8 hours. It's a IDoc to IDoc scenario with simple mapping involved.

I Have looked at various weblogs with the design optons such as BPM collect pattern, collect pattern using IDoc XML file and IDoc Packaging.

I am looking for inputs from real time experiences on such scenarios...i.e performanace statstics, best practices to handle such scenarios, hardware, XI basis tuning etc...

Please share any experiences on the same...

Regrads

Anand

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hi,

Check for IDoc tunneling and packaging

see the below links

/people/michal.krawczyk2/blog/2005/12/04/xi-idoc-bundling--the-trick-with-the-occurance-change

Also see the below link performance tuning .

Performance Tuning Checks in SAP Exchange Infrastructure (XI) - /people/prasad.illapani/blog/2007/03/08/performance-tuning-checks-in-sap-exchange-infrastructure

http://help.sap.com/bp_bpmv130/Documentation/Operation/TuningGuide.pdf

Regards

Chilla

MichalKrawczyk
Active Contributor
0 Kudos

Hi,

how can he use tunneling if he's doing a mapping ?

Regards,

michal

-


<a href="/people/michal.krawczyk2/blog/2005/06/28/xipi-faq-frequently-asked-questions"><b>XI / PI FAQ - Frequently Asked Questions</b></a>

Former Member
0 Kudos

Hi ,

Thanks Michal,

<i> I am looking for inputs from real time experiences on such scenarios...i.e performanace statstics, best practices to handle such scenarios, hardware, XI basis tuning etc...</i> And simple mapping.

his questions is see above.due to that reason i said tunneling in the performance point of view only.

Regards

Chilla

Answers (1)

Answers (1)

MichalKrawczyk
Active Contributor
0 Kudos

hi,

I believe this will not be something big for XI (don't know how big those

IDOCs are) but it will certainly be a challenage

for the receiving R3, ERP and this is the place where you should

concentrate mostly

for XI there are many ways you can go but if this will be a simple

one - one flow then you can control it in many ways

(sending 1000 IDOCs per one call etc)

but for the receiver R3 this will be something more...

Regards,

michal

-


<a href="/people/michal.krawczyk2/blog/2005/06/28/xipi-faq-frequently-asked-questions"><b>XI / PI FAQ - Frequently Asked Questions</b></a>

Former Member
0 Kudos

Idoc is a small idoc and mapping is also simple.

The options I was considering was

1. collect Idoc using idoc adapter(Event driven message processing)->BPM Collect Pattern with multi-mapping->send Idocs as package.

2. collect idoc as a idoc XML file(using idoc XML port)->file adpter pulls idoc XML files->perofrm mapping ->send idocs as pacakge.

I had look at most of the weblogs and tuning guides , but to decide on the design option I was looking for a performanance stats(time taken) and approaches followed in real time scenarios.

Any other points to be considered which can make a differenece in the performance ?

Regards

Anand

MichalKrawczyk
Active Contributor
0 Kudos

Hi,

1. for sure don't use a BPM (no collect patterns)

2. just like I said XI will not be as big problem as the receiving R3 I believe

try posting a sample of 10.000 IDOCs to your receiver R3 and check

how much time does it take to process... as you will need

at least 240 more time to process them all

Regards,

michal

Former Member
0 Kudos

HI,

No need of use BPM,

You can collect Idoc without BPM also .

Even if you have high volume of Idocs no matter, tell me exact volume or problem , so that , you will receive exact solution.

/people/stefan.grube/blog/2006/09/18/collecting-idocs-without-using-bpm - collecting IDocs without BPM

Regards

Chilla

Former Member
0 Kudos

Thanks Michal and Chandra...

Chandra,

I have looked into the weblog mentioned( option 2 - collecting idocs using IDoc XML port).

I undersatnd that performance would also depened on hardware,tuning, idoc size, mapping complexity etc, but just get a rough idea and to decide on the design I was looking for stats, approaches followed in other projects and past experiences in such scenarios.

Michal,

I cannot avoid collection of idocs as sending idoc's individually from source to target will have performance problems.

I can avoid BPM by using IDoc XML port collection ( option 2).

2.4 million idocs in 8 hours would mean that XI needs to process 300K idocs per hour. I guess I need to try option 2 and check the performance.

Regards

Anand