Integrating MDMP and Unicode systems with IDoc interfaces

Question

Hi,
We are working on integrating SAP R/3 6.20 (MDMP) with SAP PI 7.0  SP10 (Unicode) system.  The source will send MATMAS or CLFMAS IDocs with Thai / Japanese characters and PI should transform and post it to SAP ECC 5.0 Target system [IDoc to IDoc scenario].
( SAP R/3 Legacy 620 - Non-Unicode / MDMP ) |---->  ( XI  - Unicode) ->  (ECC - Unicode)
Had a look at few SAP notes (745030 ,656350 and 613389)  and it looks like there is no standard way/best practice to handle this scenario. 
References:
1.PDF of TECHED Session ID: IM101 Dealing with Multi-Language Garbage?Data – Lessons Learned
2.SAP Note 745030 - MDMP - Unicode Interface_Solution Overview.pdf
3.MDMP_Unicode_Transfer_final.doc from SAP Note 745030
4.SAP Note 656350 - Master Data Transfer UNICODE to MDMP Systems with ALE.pdf
5.SAP Note 613389 - ALE SAP system group with Unicode systems (Solution-2)
My understanding per SAP Notes: (Please correct me if I'm wrong)
a. For MDMP integraton we can't use Standard ALE Config instead we have to use Custom Configuration (IDoc collection setting in Partner profile &Scheduling  RSEOUT002/RSEOUT00_MDMP and use Function module :IDOC_INBOUND_ASYNCHRONOUS_2 ).
b. The RFC Destinations should use proper logon langauage for correct MDMP IDoc transfer (For sending IDocs with Japanese character, the logon language should be JP).
c. If we want to transfer IDocs in more than one language we need to create multiple Partner profiles/RFC destinations each with specific logon language.
Please guide us in integrating these systems if you have done a similar integrations, following are my questions:
1] Is there any configuration change required at PI layer?
2] Do we need to install codepages in PI Unicode system for all languages used or Unicode system is capable of handling all the languages?
3] Is it necessary to install any SAP Add-on package in R/3 MDMP system inorder to support MDMP to Unicode data transfer?
3] If we want to send MATMAS/CLFMAS IDoc with Thai/Japanese characters from same system, what are the changes required at the Source system?
    (Source may send MATMAS/CLFMAS IDoc with either Thai/Japanese characters but not both of them in a single IDoc)
4] Can we use regualr ALE & Partner profile settings for handling multi-byte characters or we need to use IDoc collection and RSEOUT002/RSEOUT00_MDMP report for transfer?
5] Is there any restrictions on the IDoc types (MATMAS,CLFMAS etc) supported in MDMP-Unicode integration solution.
5] Is there any best practice document available for this scenario?
6] Do we need to involve SAP AG for MDMP to Unicode system integrations(As per SAP Note: 656350) ?
Thanks and Regards,
Ananth

Answer 1

Hi Ananth,
as you have already mentioned, you need differents RFC destinations for each language. So you have to make sure, that the IDocs use the right destination according to there content.
If you have messages from PI to MDMP it is the same, you need different channels with different logon languages as well. You need an identifier in the message, that can be used for selecting the correct channel.
It should not be a restricting to any IDoc type, but it is not possible to post a message with different languages (which require different codepages) in one IDoc.
For correct conversion from a non unicode system to unicode, the codepages have to be installed in th OS of the PI server.
Regards
Stefan