Hello Experts,
I am working on one performance item and I have already applied some changes to the original version.
Now, If I compare my new program with old program, I have good improvement in performance. I am checking if I can do anything on statements that are top on the below list. I think 40% for Modify statement is acceptable after my research (Below run is updating around 20M records which is real time volume for this application).
As we can see 34% of run time to going for one SELECT query on custom table. Take a look at below high level flow of my program to understand above select query.
1. Select data from ZABC
2. Select data from Variant Table (Var1, Var2, Var3 etc.., 12 in real time)
3. Loop Variant Table
4. Select data from X, Y, Z table for Var<n>.
5. Populate final internal table from ZABC, X, Y and Z table
6. Modify ZTABLE with Final Internal table data
7. End Loop on Variant Table
As described in the flow of the program, ZABC table data is common for all the variants and need not to fetch multiple times. Hence I am doing it only once in my program. Below is that select query:
select rrcty ryear rbukrs racct rcntr sum( amt1) as amt1 "Like I have 32 amount fields in original query from zabc into table i_zabc where ryear in r_year " Two records in ranges with I and EQ and rvers = '001' and rrcty in r_rrcty "Three records in ranges with I and EQ and rldnr = 'DT' group by rrcty ryear rbukrs racct rcntr order by rrcty ryear rbukrs racct rcntr.
Now, third statement in my trace results, with 10% of overall time is this:
loop at i_abc assigning <fs_abc>.
...
loop at i_table assigning <fs_table> where low <= <fs_abc>-racct and high >= <fs_abc>-racct.
...
endloop.
...
endloop.
6 million executions with this complex WHERE condition is causing this statement to get 3rd position in trace results. I tried below two options which are, I think, taking even more time - (I am still monitoring these options)
1) Removed WHERE condition on LOW, HIGH and applied filter inside the loop.
2) Removed WHERE condition on HIGH only and applied filter inside the loop.
Any suggestions on how to proceed with ZABC selection and I_TABLE loop.
Let me know if you have any questions on above compose.