Select from internal table or Filter internal table

[dhawal_patil]

I want to filter an internal table and store the data into another internal table based on a condition. I know of two ways, I can achieve it, i.e. with "SELECT from @itab" and "FILTER #( itab WHERE field1 = 'SOME_VALUE' )".

I want to know if there is any difference between the performance of these two approaches? Is any one of them preferred more than another?

Here's a sample code of what I am trying to do:

TYPES: BEGIN OF ty_itab,
        field1 TYPE char15,
        field2 TYPE char15,
       END OF TY_GLINTDATA.

DATA: itab TYPE SORTED TABLE OF ty_itab WITH NON-UNIQUE KEY field1.


SELECT field1, field2 from ztable into table @itab.

"Using SELECT from @itab
SELECT field1, field2 
  FROM @itab as a
  WHERE field1 <> 'SOME_VALUE'
  INTO TABLE @DATA(itab_filtered).
                 

"Using filter
DATA(itab_filtered) = FILTER #( itab where field1 <> 'SOME_VALUE' ).

[Sandra_Rossi · Accepted solution]

As rule-of-thumb, execution of simple logic on application server will always be faster than execution on database system, because you avoid network/data transfer.

[dhawal_patil · Accepted solution]

Thanks, Sandra. From the documentation of SELECT - FROM @itab I understand that the data is not transported to database system if the SELECT statement meets the requirements for table buffering. If that is the case with my sample code in the question, won't the SELECT statement in my code be executed on application server and avoid network/data transfer?

Correct me if my understanding is wrong here. I am just curious about how the "SELECT - FROM @itab" works and in which cases it can be used?

[Sandra_Rossi · Accepted solution]

You understand well. But few tables are buffered, so, as rule-of-thumb... (as I said)

Better write natively-optimized code than relying on hypothetical same-performance SQL.

Now, it takes few minutes to test the performance by yourself...