Nested Loop Vs Parallel Cursor Technique

Scenario: Many times In development scenario we come across situations where we need to have a nested loop to find our the matching records in the Header and the Item table. In that case we follow Nested loop which takes a longer processing time. To Improve the program performance the Nested Loop Condition can be optimized in a certain way called Parallel Cursor technique to improve the performance.

The below post demonstrates the same by considering the two table : BKPF & BSEG.

_________________________________________________________________________________

REPORT  zparallel_cursor_test.

DATA : lt_bkpf TYPE TABLE OF bkpf,
              lt_bseg TYPE TABLE OF bseg,
              ls_bkpf TYPE bkpf,
              ls_bseg TYPE bseg.
DATA : lv_time1 TYPE i,
              lv_time2 TYPE i,
              lv_time3 TYPE i,
              lv_time4 TYPE i,
              lv_res_time_gn_process TYPE i,
              lv_res_time_pa_process TYPE i,
              indx TYPE i.
DATA : lv_bkpf_recs TYPE i.
DATA : lv_bseg_recs TYPE i.

START-OF-SELECTION.

  SELECT * FROM bkpf INTO TABLE lt_bkpf.
  SELECT * FROM bseg INTO TABLE lt_bseg.

  DESCRIBE TABLE lt_bkpf LINES lv_bkpf_recs.
  DESCRIBE TABLE lt_bseg LINES lv_bseg_recs.

  GET RUN TIME FIELD lv_time1.
  LOOP AT lt_bkpf INTO ls_bkpf.
    LOOP AT lt_bseg INTO ls_bseg WHERE vbeln = ls_bkpf-belnr.
* BUSINESS LOGIC GOES HERE.
    ENDLOOP.
  ENDLOOP.

  GET RUN TIME FIELD lv_time2.

  lv_res_time_gn_process = lv_time2 - lv_time1.

  * Applying Parallel Cursor Technique

 SORT lt_bkpf BY belnr.
  SORT lt_bseg BY belnr.

  GET RUN TIME FIELD lv_time3.
  LOOP AT lt_bkpf INTO ls_bkpf.
    READ TABLE lt_bseg INTO ls_bseg WITH  KEY belnr = ls_bkpf-belnr BINARY SEARCH.
    IF sy-subrc = 0.
      indx = sy-tabix.
      LOOP AT lt_bseg INTO ls_bseg FROM indx.
        IF ls_bseg-belnr <> ls_bkpf-belnr.
          EXIT.
        ENDIF.
* BUSINESS LOGIC GOES HERE.
      ENDLOOP.
    ENDIF.
  ENDLOOP.

  GET RUN TIME FIELD lv_time4.

  lv_res_time_pa_process = lv_time4 - lv_time3.

  WRITE :/ 'Comparision Between Nested Loop And Parallel Cursor Technique'.
  WRITE:/ 'No Of records in BKPF(Header) Table = ', lv_bkpf_recs.
  WRITE:/ 'No Of records in BSEG(Item) Table = ', lv_bseg_recs.

  Write :/ 'Time Taken in Normal Nested loop Processing', lv_res_time_gn_process  ,'Vs.' ,
           ' Time Taken In Parallel Cursor Technique', lv_res_time_pa_process.
_________________________________________________________________________________

Output:
_________________________________________________________________________________

_________________________________________________________________________________