When I was responsible for CRM Fiori application, I once meet with a performance issue.
When the users perform the synchronization for the first time on their mobile device, the opportunities belonging to them will be downloaded to mobile which is so called the initial load phase. The downloaded data includes attachment header information.
Since for Attachment read in CRM, there is no multiple-enabled API, so we have to perform the single read API within the LOOP, which means the read is performed sequentially:
We really suffer from this poor performance.

As explained in my blog What should an ABAPer continue to learn as an application developer,

I get inspiration from the concept Parallel computing

which is usually related to functional programming language. A function which has no side-effect, only manipulates with immutable data set is a good candidate to be handled concurrently. When looking back on my performance issue, the requirement to read opportunity attachment header data perfectly fits the criteria: only read access on header data, each read is segregated from others – no side effect. As a result it is worth a try to rewrite the read implementation into a parallelism version.
The idea is simple: split the opportunities to be read into different parts, and spawn new ABAP sessions via keyword STARTING NEW TASK, each session is responsible for a dedicated part.
The screenshot below is an example that totally 100 opportunities are divided into 5 sub groups, which will be handled by 5 ABAP sessions, each session reads 100 / 5 = 20 opportunity attachments.
The parallel read version:
    DATA:lv_taskid        TYPE c LENGTH 8,
         lv_index         TYPE c LENGTH 4,
         lv_current_index TYPE int4,
         lt_task          LIKE it_orders,
         lt_attachment    TYPE crmt_odata_task_attachmentt.

* TODO: validation on iv_process_num and lines( it_orders )
    DATA(lv_total) = lines( it_orders ).
    DATA(lv_additional) = lv_total MOD iv_block_size.
    DATA(lv_task_num) = lv_total DIV iv_block_size.
    IF lv_additional <> 0.
       lv_task_num = lv_task_num + 1.
    DO lv_task_num TIMES.
      CLEAR: lt_task.
      lv_current_index = 1 +  iv_block_size * ( sy-index - 1 ).
      DO iv_block_size TIMES.
        READ TABLE it_orders ASSIGNING FIELD-SYMBOL(<task>) INDEX lv_current_index.
        IF sy-subrc = 0.
          MOVE-CORRESPONDING <task> TO <cur_task>.
          lv_current_index = lv_current_index + 1.

      IF lt_task IS NOT INITIAL.
        lv_index = sy-index.
        lv_taskid = 'Task' && lv_index.
          STARTING NEW TASK lv_taskid
          CALLING read_finished ON END OF TASK
            it_objects = lt_task.

    WAIT UNTIL mv_finished = lv_task_num.

    rt_attachments = mt_attachment_result.

    DATA: lt_attachment TYPE crmt_odata_task_attachmentt.

    ADD 1 TO mv_finished.
      ct_attachments              = lt_attachment
      system_failure        = 1
      communication_failure = 2.

    APPEND LINES OF lt_attachment TO mt_attachment_result.

In function module ZJERRYGET_ATTACHMENTS, I still use the attachment single read API:
*"*"Local Interface:

DATA(lo_tool) = new zcl_crm_attachment_tool( ).

ct_attachments = lo_tool->get_attachments_origin( it_objects ).

So in fact I didn’t spend any effort to optimize the single read API. Instead, I call it in parallel. Let’s see if there is any performance improvement.
In this test report, first I generate an internal table with 100 entries which are opportunity guids. Then I perform the attachment read twice, one done in parallel and the other done sequentially. Both result are compared in method compare_read_result to ensure there is no function loss in the parallel version.

Testing result ( unit: second )

It clearly shows that the performance increases with the number of running ABAP sessions which handles with the attachment read. When the block size = 100, the parallel solution degrades to the sequential one – even worse due to the overhead of WAIT.
( for raw test result screenshot, you can find it here )
For sure in productive usage the number of block size should not be hard coded.
In fact in my test code why I use the variable name iv_block_size is to express my respect to the block size customizing in tcode R3AC1 in CRM middleware 🙂
All related source code could be found from my github.
To report this post you need to login first.


You must be Logged on to comment or reply to a post.

  1. Spencer Liang

    This is an inspiring page. Thanks for sharing.

    Performance issue might be complicated in application area and especially on production system.

    Your example may theoretically improve data processing but application is also restricted by system resources such work process, rfc source, memory consumption, CPU, etc.

    I am wondering if you could add some input about these thresholds.


    1. Jerry Wang Post author
      Hello Spencer,
      Nice to see your comment here. Your blog Search “Belonging To” = My Team & My Group and how it works in CRM


      is very helpful for me When I was responsible for development on CRM Fiori application “My Opportunity”. Thank you 🙂
      Regarding your comment, yes you are right, the performance enhancement is available from theoretical point of view when handling logic is switched to be done in parallel, however in real case if there is high workload in customer server which means there is not enough work process to finish the task in parallel, then the degree to this performance gain is suspectable.
      We finally did not move my prototype into productive code, after an effort estimation if we need to deliver a switch which allows customer to toggle between sequential and concurrent style just the same as standard Gateway handling for OData batch request.
      The estimated effort is high and we have other higher priority task to do. In the end the performance improvement is done based on the API code in more finer granularity.
      Best regards,
  2. Abdul Hakim

    Hi Jerry,

    Thanks for sharing this useful blog with the ABAP community.

    I don’t see the error handling in your program for instance when the resources are completely occupied how you are handling it. Also you have not used DESTINATION IN GROUP and hence your new task will always run on the current application server which may lead to occupying other process. I think this is not a good idea. Please clarify.





    1. Jerry Wang Post author

      Hi Hakim,

      Thanks a lot for your comment, yes this is just a prototype so no error handling is available. And I didn’t know the option DESTINATION IN GROUP you mentioned. Thank you very much for pointing this out so that I learned another new stuff today 🙂

      Best regards,



Leave a Reply