In a lot of scenarios we come across where nesting loops is the only option we have to achieve what we want.
They can be hazardous if not tuned properly and can result in time out dumps and heavy processing time.
The solution : PARALLEL Cursor.
Lets assume a scenario where we are having 3 internal tables one with following structure.
KNA1
Kunnr
Name1
VBRK
Vbeln
Kunag [Sold to will be equal to kunnr].
VBRP
vbeln
posnr
matnr
netwr
Now it becomes tricky as we want to link KNA1 to VBRK and VBRK to VBRP.
Note : This is just to explain a three level nested loop if exact case is considered a read on KNA1 will be better.
Some thumb rules for Parallel Cursor
1. SORTING should be on the field which will part of where clause if you are not using.
Like in this case KNA1 will be on KUNNR
VBRK will be on KUNAG
and VBRP will be on VBELN.
2. Recommended : use of variables to store tabix of the loop process.
Now the solution :
DATA : lv_vbrk_tabix TYPE sy-tabix,
lv_vbrp_tabix TYPE sy-tabix.
SORT it_kna1 BY kunnr.
SORT it_vbrk BY kunag.
SORT it_vbrp BY vbeln.
lv_vbrk_tabix = 1.
LOOP AT it_kna1 INTO wa_kna1.
WRITE:/5 'Name : ' , wa_kna1-name1 , ' | ' , wa_kna1-kunnr.
ULINE.
LOOP AT it_vbrk INTO wa_vbrk FROM lv_vbrk_tabix.
IF wa_kna1-kunnr GT wa_vbrk-kunag.
lv_vbrk_tabix = sy-tabix.
EXIT.
ENDIF.
CHECK wa_kna1-kunnr = wa_vbrk-kunag.
WRITE:/10 wa_vbrk-vbeln , wa_vbrk-fkdat.
READ TABLE it_vbrp WITH KEY vbeln = wa_vbrk-vbeln
BINARY SEARCH
TRANSPORTING NO FIELDS.
IF sy-subrc IS INITIAL.
lv_vbrp_tabix = sy-tabix.
ELSE.
CONTINUE.
ENDIF.
LOOP AT it_vbrp INTO wa_vbrp FROM lv_vbrp_tabix.
IF wa_vbrk-vbeln NE wa_vbrp-vbeln.
EXIT.
ENDIF.
WRITE:/15 wa_vbrp-matnr , 50 wa_vbrp-netwr.
ENDLOOP.
ENDLOOP.
ENDLOOP.
These are the 2 possible ways of parallel cursor.
Do comment if any issues or there is a better way... 