Custom extractor performance issues

[Former Member]

We have 1 custom extractor to pull Shipments and Deliveries data from VTTK, VTTP, LIKP, LIPS tables. There are reasons why went with custom instead of standard one.

Problem is with performance. We are trying to pull shipments and deliveries which are changed

in last 3 days or created in last 3 days. The extractor is damn slow. How do we improve performance.

1 option is to create indexes on created on and changed on date so that selects on tables is faster.

Is there any other option?

[snitwipro]

Hi,

   It is very difficult to anticipate the problem without knowing the technical design. However, please try SQL trace and then try a test extraction. Please see if you can make the extractor delta enable so that you can fetch changed records rather than extracting everything. Also please check if you have the correct joins.

Regards

Sourav

[former_member182516]

In Addition to Hendrik post...  please post the below details.

What is the extraction method you are using ? i mean -- FM based ot View based?

Could you post the code that you had written if you are using FM based extraction?

Try to debug and fine tune the code with the help of ABAP person.

If the extractor is build on the view --> you need to fine tune the view... check the time in ECC if you are able to get the data displayed. Also try to execute RSA3 and check the time of extraction.

Check in SM66 --> where the PID is getting stuck while executing the Info package (you need to monitor the PID in ECC system so that you can come to know where the process is really getting stuck).

Regards

KP

[RamanKorrapati]

Hi Vishnu,

Please check below points.

1. if you have any enhancement, try to use SQL trace

2. As said above, during your data load, look at T - Code SM50/51

3. Check at SM58(Luws may stuck)

4. is this happens today only or daily

5. At Which step this load is slow, before PSA?

6. BW Application server may be busy during your load

Thanks

[hendrik_brandes]

Hello Vishnu,

without knowing exactly what your extractor do, I try to give you some more options:

1) Do you try to move as many as possible operations to the DB? Meaning: Using SQL-Join, Grouping, Views ...

2) Use indices will improve your speed if your SQL-statement will select the proper fields in the where-clause.

3) Use SQL-cursor for a package-wise data selection. This will improve the speed within the process due to smaller packages.

4) Have you checked your ABAP-coding tuned? Use Field-Symbols in LOOPs, try to use Hashed-Tables for reading operations on internal tables.

5) Select only those fields, which are really necessary and do not use INTO CORRESPONDING FIELDS OF TABLE

Kind regards,

Hendrik