posting document by BAPI vs BDC using RFC

[Former Member]

We change to use BAPI update via RFC from legacy system instead of BDC. i.e. BAPI_ACC_DOCUMENT_POST and BAPI_ACC_DOCUMENT_CHECK.

In test environment, the performance is improved. Only 1/2 original time. However, it is no improvement in PRD. Is there any configuration or parameter I can further check in PRD (either in BAPI or RFC) ?

thanks a lot,

Doris

Edited by: Doris Kam on Apr 9, 2009 4:35 AM

[Former Member]

Hi,

1. How do you place your commit work statement or BAPI_COMMIT_TRANSACTION in the program? Do you commit it for every transaction or a group of transactions? SAP recommends to issue commit command for every 100 up to 1000 transactions. To be safe, you can use it for every 500 transactions.

2. What about your other ABAP statement? Do you have nested LOOP, nested SELECT? Could you use field symbols instead of work area in looping the internal table?

Hope it helps.

Regards,

Lim...

[Former Member]

Hi Lim,

Thanks for your prompt reply.

1. We use COMMIT WORK after BAPI_ACC_DOCUMENT_POST. Only one FI document will be included when the program is being called. We call RFC again if more than one document.

2. We use two LOOP only, read header and then detail records.

In my case, document contains average 500 line items and it takes 6+ minutes to complete one RFC. However, it takes less than one minutes for a document with 800 line items in test environment. Wonder why so big difference.

Doris

[Former Member]

Appreciated if anyone can provide any other helpful info.

[Former Member]

Hi Doris,

Go to your test environment, and do:

1. SE30, and try to check the performance analysis result

2. ST05, and try to see which SQL statement that caused the issue

From this 2 tools, you might get different view, like it may be not the BAPI that is costly but the select or loop statements elsewhere in the program. It's hard to tell you if we don't see the trace or log result.

Regards,

Lim...

[ThomasZloch]

You should try to find out where the time is spent, actual processing or RFC communication. What if a document is posted directly in the remote systems (not via RFC), is the processing time the same in test and production? If yes, then there might be a network problem or bottleneck of some sort.

There is also an "RFC Trace" in SAP, which I have not used in practice yet, so maybe you can google up some info or wait for other replies here.

Thomas