topic Re: Performance issue in Application Development and Automation Discussions

Performance issue

Former Member — Sat, 15 Nov 2008 04:53:07 GMT

Moved by moderator to the correct forum. Please use a more meaningful subject in future.

Hi All,

Here below the code the Loop under Read statement i have used and have to use the binary search where have to use can any one suggest me to execute it fast .

SORT gvt_item   BY vbeln posnr.
  Read table gvt_item with key vbeln = ls_header-vbeln
                             posnr = ls_header-posnr.
   if sy-subrc = 0.
      wa_index = sy-tabix.

      gvt_item-zz_crm_id = ls_header-zz_crm_id.
      gvt_item-ktext     = ls_header-ktext.
      gvt_item-faksk     = ls_header-faksk.
      gvt_item-fpbez     = ls_header-fpbez.
      gvt_item-bezeich   = ls_header-bezeich.
      gvt_item-faksp     = ls_header-faksp.
      gvt_item-anndat    = ls_header-anndat.
      gvt_item-candat    = ls_header-candat.
      gvt_item-stats     = ls_header-stats.
      gvt_item-autte     = ls_header-autte.
      gvt_item-cmana     = ls_header-cmana.
      gvt_item-ename     = ls_header-ename.
      gvt_item-pstyv     = ls_header-pstyv.
      gvt_item-fpart     = ls_header-fpart.

* Insert by BRDK933549, modify the anniversary logic

* End of BRDK933549

      MODIFY gvt_item INDEX sy-tabix
             TRANSPORTING zz_crm_id ktext faksk fpbez bezeich
              faksp anndat candat stats autte cmana ename pstyv fpart.

    Endif.
*    ENDLOOP.

Thanks in Advance

Edited by: Matt on Nov 16, 2008 8:22 AM Don't forget to put tages around your code to keep the formatting

Re: Performance issue

Former Member — Sat, 15 Nov 2008 05:03:43 GMT

Do F1 on binary search.

And post thread at their relevant place.

Re: Performance issue

Former Member — Sat, 15 Nov 2008 21:04:38 GMT

Hi,

SORT gvt_item BY vbeln posnr. .

loop XXXXXX.

Read table gvt_item with key vbeln = ls_header-vbeln

posnr = ls_header-posnr

binary search.

endloop.

Thanks,

Sree,

Re: Performance issue

matt — Sun, 16 Nov 2008 07:37:04 GMT

The absolute fastest way, if you've a reasonable amount of data is to us ASSIGNING for your read - you can change the data directly in the table, and a HASHED table - no BINARY SEARCH required.

Assuming your data is of type "table_type". First, define your table to be HASHED.

DATA: gvt_time TYPE HASHED TABLE OF table_type WITH NON-UNIQUE KEY vbeln posnr.

FIELD-SYMBOLS: <ls_wa> TYPE LINE OF table_type.

then


  Read table gvt_item with TABLE key vbeln = ls_header-vbeln
                             posnr = ls_header-posnr ASSIGNING <ls_wa>.
   if sy-subrc = 0.
"     wa_index = sy-tabix.
 
      <ls_wa>-zz_crm_id = ls_header-zz_crm_id.
      <ls_wa>-ktext     = ls_header-ktext.
      <ls_wa>-faksk     = ls_header-faksk.
      <ls_wa>-fpbez     = ls_header-fpbez.
      <ls_wa>-bezeich   = ls_header-bezeich.
      <ls_wa>-faksp     = ls_header-faksp.
      <ls_wa>-anndat    = ls_header-anndat.
      <ls_wa>-candat    = ls_header-candat.
      <ls_wa>-stats     = ls_header-stats.
      <ls_wa>-autte     = ls_header-autte.
      <ls_wa>-cmana     = ls_header-cmana.
      <ls_wa>-ename     = ls_header-ename.
      <ls_wa>-pstyv     = ls_header-pstyv.
      <ls_wa>-fpart     = ls_header-fpart.
 
* Insert by BRDK933549, modify the anniversary logic
 
* End of BRDK933549
 
"     MODIFY gvt_item INDEX sy-tabix
"           TRANSPORTING zz_crm_id ktext faksk fpbez bezeich
"           faksp anndat candat stats autte cmana ename pstyv fpart.
 
    Endif.
"    ENDLOOP.

Note, the use of tables with header-lines is bad programming practice. Avoid where possible.

matt

Re: Performance issue

Former Member — Tue, 18 Nov 2008 04:00:43 GMT

Hi,

Just place binary search after read stmt.

remember when soring the table before reading it, use 'binary search' for fast execution.


  Read table gvt_item with key vbeln = ls_header-vbeln
                             posnr = ls_header-posnr binary search.

Tks,

Krishna

Re: Performance issue

Former Member — Wed, 26 Nov 2008 12:43:13 GMT