2011 Jul 25 12:25 PM
Hello,
I am going to create project on lsmw for data migration from legacy system. In this specific case it is not just migration of the data to SAP system but its processing the data an dthen creating objects in SAP system. Hence its very perfomarmace critical.
I amn using Idoc and BAPI in lsmw.
My question how to improve performance here .
My options:
1. Call the API(which is processing the data) in side the BAPI with new task for each independent set.
2. No need of (1) if I am doing parallel processing in step idoc processing. Here lsmw itself provides parallel processing of Idoc.
3. Use FPP . but since FPP can be integrated only for data in database, I would need to first store the incoming data in a temp tables and then call API in FPP run where API reads data from temp tables .
Can you please suggest from your experience which is the best technique to improve the perfromance in this situation ?
2011 Jul 25 1:18 PM
The first option is the best - with each BAPI call in an asynchronous RFC.
The second option is also good - no need to write any parallel processing code - but there is a gateway limitation on number of IDocs being posted per minute.
The third option is not elegant one as it saves data in temporary tables.
Regards
Suresh
2011 Jul 25 1:18 PM
The first option is the best - with each BAPI call in an asynchronous RFC.
The second option is also good - no need to write any parallel processing code - but there is a gateway limitation on number of IDocs being posted per minute.
The third option is not elegant one as it saves data in temporary tables.
Regards
Suresh
2011 Jul 26 6:08 AM
Hello Suresh,
Thank you so much for resply.
My proiblem with lsmw is that I cannot call API in background job but it wil run in dialog process.
That was one reason to think of storing it temporarily and tehn processing it in FPP.
But just today I came across what is called as data transfer workbench help document where it says that you can process job in background . Can anyone confirm this ?
If yes then my next question is how to enable parallel processing using workbench