Performance problem in smartform(abap)

[Former Member]

HI floks

i did some development in new smartform its working fine but i have issue in data base performance is 76% . but i utilize similar below code with various conditions in various 12 places . is it possible to reduce performance this type of code . check it and mail me how can i do it . if possible can suggest me fast .how much % is best for this type of performance issues.

DATA : BEGIN OF ITVBRPC OCCURS 0,

LV_POSNR LIKE VBRP-POSNR,

END OF ITVBRPC.

DATA : BEGIN OF ITKONVC OCCURS 0,

LV_KNUMH LIKE KONV-KNUMH,

LV_KSCHL LIKE KONV-KSCHL,

END OF ITKONVC.

DATA: BEGIN OF ITKONHC OCCURS 0,

LV_KNUMH LIKE KONH-KNUMH,

LV_KSCHL LIKE KONH-KSCHL,

LV_KZUST LIKE KONH-KZUST,

END OF ITKONHC.

DATA: BEGIN OF ITKONVC1 OCCURS 0,

LV_KWERT LIKE KONV-KWERT,

END OF ITKONVC1.

DATA : BEGIN OF ITCALCC OCCURS 0,

LV_KWERT LIKE KONV-KWERT,

END OF ITCALCC.

DATA: COUNTC(3) TYPE n,

TOTALC LIKE KONV-KWERT.

SELECT POSNR FROM VBRP INTO ITVBRPC

WHERE VBELN = INV_HEADER-VBELN AND ARKTX = WA_INVDATA-ARKTX .

APPEND ITVBRPC.

ENDSELECT.

LOOP AT ITVBRPC.

SELECT KNUMH KSCHL FROM KONV INTO ITKONVC WHERE KNUMV =

LV_VBRK-KNUMV AND KPOSN = ITVBRPC-LV_POSNR AND KSCHL = 'ZLAC'.

APPEND ITKONVC.

ENDSELECT.

ENDLOOP.

SORT ITKONVC BY LV_KNUMH.

DELETE ADJACENT DUPLICATES FROM ITKONVC.

LOOP AT ITKONVC.

SELECT KNUMH KSCHL KZUST FROM KONH INTO ITKONHC WHERE KNUMH = ITKONVC-LV_KNUMH AND KSCHL = 'ZLAC' AND KZUST = 'Z02'.

APPEND ITKONHC.

ENDSELECT.

ENDLOOP.

LOOP AT ITKONHC.

SELECT KWERT FROM KONV INTO ITKONVC1 WHERE KNUMH = ITKONHC-LV_KNUMH AND

KSCHL = ITKONHC-LV_KSCHL AND KNUMV = LV_VBRK-KNUMV.

MOVE ITKONVC1-LV_KWERT TO ITCALCC-LV_KWERT.

APPEND ITCALCC.

ENDSELECT.

endloop.

LOOP AT ITCALCC.

COUNTC = COUNTC + 1.

TOTALC = TOTALC + ITCALCC-LV_KWERT.

ENDLOOP.

MOVE ITKONHC-LV_KSCHL TO LV_CKSCHL.

MOVE TOTALC TO LV_CKWERT.

it's urgent ..........

thanks .

bbbbye

suresh

[Former Member]

Hi

U can use for all entries instead of to insert the select statament into loop:

SELECT POSNR FROM VBRP INTO TABLE ITVBRPC
            WHERE VBELN = INV_HEADER-VBELN 
              AND ARKTX = WA_INVDATA-ARKTX .

IF SY-SUBRC = 0.
  SELECT KNUMH KSCHL FROM KONV INTO TABLE ITKONVC
    FOR ALL ENTRIES IN  ITVBRPC 
         WHERE KNUMV = LV_VBRK-KNUMV 
           AND KPOSN = ITVBRPC-LV_POSNR 
           AND KSCHL = 'ZLAC'.
  
  SELECT KNUMH KSCHL KZUST FROM KONH INTO TABLE ITKONHC
          FOR ALL ENTRIES ITKONVC
                WHERE KNUMH = ITKONVC-LV_KNUMH 
                  AND KSCHL = 'ZLAC' AND KZUST = 'Z02'.
  IF SY-SUBRC = 0.
   SELECT KWERT FROM KONV INTO TABLE ITCALCC 
          FOR ALL ENTRIES ITKONHC
                  WHERE KNUMV = LV_VBRK-KNUMV
                    AND KSCHL = ITKONHC-LV_KSCHL 
                    AND KNUMH = ITKONHC-LV_KNUMH. 

   ENDIF.
 ENDIF.

Max

[Former Member]

I think you'll find that if you follow Max's example, the database percentage will actually go up!

But this is because his code is more efficient on the non-database portion of the program. It will eliminate a lot of network traffic between the database and the application

My point is that looking at the database percentage doesn't really tell you much. A program that retrieves a lot of data but doesn't manipulate it will naturally have a high database percentage. One that doesn't get much data but processes heavily what it does get will have a lower database percentage.

The important things to keep in mind are to use an index (which you are already doing) and avoid nested loops (which doesn't apply here).

Rob

[Former Member]

u can use INTO TABLE instead of INTO & ENDSELECTION.

u can use FOR ALL ENTRIES instead of LOOPS & NESTED LOOPS.

REGDS

KIRAN