A simple performance comparison against different types of internal tables
I construct three internal tables with different table types:
The complete test source code could be found in the end part of the blog.
insert operation comparison
The hashed table is least efficient since additional overhead is paid to maintain the internal administrative information for hash logic.
The standard table is fastest due to the fact that there is no overhead.
read operation comparison
The standard table read is slowest due to o(n) complexity.
If we exclude the standard table read and compare the left three, it is clear the hashed table read is most efficient.
The complete test source code:
REPORT z. PARAMETERS: count TYPE i OBLIGATORY DEFAULT 1000. TYPES: BEGIN OF ty_pair, key TYPE i, value TYPE string, END OF ty_pair. TYPES: tt_standard TYPE STANDARD TABLE OF ty_pair WITH KEY key, tt_sorted TYPE SORTED TABLE OF ty_pair WITH UNIQUE KEY key, tt_hashed TYPE HASHED TABLE OF ty_pair WITH UNIQUE KEY key. DATA: lv_start TYPE i, lv_end TYPE i, lt_standard TYPE tt_standard, lt_sorted TYPE tt_sorted, lt_hashed TYPE tt_hashed, lv_size TYPE i. START-OF-SELECTION. lv_size = count. PERFORM insert_standard. PERFORM insert_sorted. PERFORM insert_hashed. PERFORM read_standard. PERFORM read_standard_binary. PERFORM read_sorted. PERFORM read_hashed. FORM insert_standard. PERFORM start_timer. DO lv_size TIMES. DATA(line) = VALUE ty_pair( key = sy-index value = sy-index ). APPEND line TO lt_standard. ENDDO. PERFORM stop_timer. " WRITE: / 'standard table insertion: ' , lv_end. ENDFORM. FORM insert_sorted. PERFORM start_timer. DO lv_size TIMES. DATA(line) = VALUE ty_pair( key = sy-index value = sy-index ). INSERT line INTO TABLE lt_sorted. ENDDO. PERFORM stop_timer. " WRITE: / 'sorted table insertion: ' , lv_end. ENDFORM. FORM insert_hashed. PERFORM start_timer. DO lv_size TIMES. DATA(line) = VALUE ty_pair( key = sy-index value = sy-index ). INSERT line INTO TABLE lt_hashed. ENDDO. PERFORM stop_timer. " WRITE: / 'hashed table insertion: ' , lv_end. ENDFORM. FORM read_standard. PERFORM start_timer. DO lv_size TIMES. READ TABLE lt_standard ASSIGNING FIELD-SYMBOL() WITH KEY key = sy-index. ASSERT sy-subrc = 0. ENDDO. PERFORM stop_timer. WRITE:/ 'standard table read: ', lv_end. ENDFORM. FORM read_standard_binary. SORT lt_standard BY key. PERFORM start_timer. DO lv_size TIMES. READ TABLE lt_standard ASSIGNING FIELD-SYMBOL() WITH KEY key = sy-index BINARY SEARCH. ASSERT sy-subrc = 0. ENDDO. PERFORM stop_timer. WRITE:/ 'standard table binary read: ', lv_end. ENDFORM. FORM read_sorted. PERFORM start_timer. DO lv_size TIMES. READ TABLE lt_sorted ASSIGNING FIELD-SYMBOL() WITH KEY key = sy-index. ASSERT sy-subrc = 0. ENDDO. PERFORM stop_timer. WRITE:/ 'sorted table read: ', lv_end. ENDFORM. FORM read_hashed. PERFORM start_timer. DO lv_size TIMES. READ TABLE lt_hashed ASSIGNING FIELD-SYMBOL() WITH TABLE KEY key = sy-index. ASSERT sy-subrc = 0. ENDDO. PERFORM stop_timer. WRITE:/ 'hashed table read: ', lv_end. ENDFORM. FORM start_timer. CLEAR: lv_start, lv_end. GET RUN TIME FIELD lv_start. ENDFORM. FORM stop_timer. GET RUN TIME FIELD lv_end. lv_end = lv_end - lv_start. ENDFORM.
New NetWeaver Information at SAP.com
Very Helpfull