How to decrease loading time for 0Mat_Plant

[Former Member]

Hi Experts,

I am currently looking to decrease loading time of 0Mat_Plant. I have gone thru performance material available and checked following things listed below :

Plz note following:

1) Standard delta is avail but it has been enhanced and there are Z fields. So if i use delta, them I might be missing delta for Field.

2)Can we switch OFF number range SID buffering to avoid memory allocation on daily basis ...somehow through ABAP program?

3) BUffering can be switched OFF, but there is delta load also alonwith full load. So performance would decrease for delta records. But no of records for Full load is 188000 and for delta is just 1700 around. SO will this helps in decreasing overall time? Any idea?

Load is taking around 90 minutes in my system (Full Load)..

I m not clear abt data package size and if it will help me or not..

What more can we do??

Any help would be appreciated and points would be assigned......

Thanks in advance

Gaurav

Message was edited by:

Gaurav

[Former Member]

Gaurav,

1. <u>Please let us why are you doing Delta and Full Load?</u>

2. Packet size will help upto some extent.

3. Why don't use new Datasource for Enhanced Fields

4. Mat Plant has lot of fields... try to remove unnecessary fields if you don't want. Sit with Business Users, ask which fields required. Ask them to predict for future as well. Remove unnecessary Fields.

5. Spliting Mat Plant into 2 Datasources is not a bad Option. In Our earlier Project we splited 0MAT_PLANT DS into 2(mat plant 52 Million records). But, in your case it is very less.

6. while loading, ask your DBA people to analyze the load. They can tell you want is going on in the DB Level.

7. How many Navigational Attributes? Determining SID values for Nav attributes could be problem.

Nagesh Ganisetti.

• assign points if it helps.

[Former Member]

Hi Nagesh,

We are doing Full load in master data chain which runs at start of batch, ater tht there were some SID's failures ( because of new ceation of transaction for whihc master was not loaded in master data chain, they were created afterwards) , so we started doing delta in those chains.

There were some z fields, for which standard delta will not take care of, so we are doing Full load.

What exactly you mean by spliting DS?? you mean running 2 infopackages?

We have already analysed as per BAsis/DBA perspective - now we need to do on our end.

Thanks for your answer.

I hve assigned points

Gaurav

[Former Member]

Spliting DS mens... use Standard DS as it is...and enhanced fields in another fields.

1. is the loading time you mentioned for all the master data or only 0mat_plant?

2. If it 0mat_plant, did you check SAP R/3 job how much time it is taking? If R/3 Job (Created because of BW Load) finished Quickly, No Problem with DS and Enhanced Fields. If it is taking extra time to complete R/3 Job, we need to work to improve the performence. Check Enhanced Code for populating data for New fields.

3. If R/3 Jobs finished quickly, check TRFC locked or not (SM58 for Background User)?

4. If R/3 Jobs finished quickly, TRFC also released. Check your coding if you have any UR or TR.

Nagesh Ganisetti.

• Assign points if it helps.

[Former Member]

Hi Nagesh,

Here are my replies;

1) Loading time is only for 0mat_plant..

2) R/3 job finishes quickly....no prob seems to be with this..

3) I m not clear about "locked trfc..but there are no errors in trfc...and there doe snot seem to be IDOC clogging problem...

4) Not sure what yu mean by trfc "released"....

I checked R/3 JOb log and its pasted below ( though for other datasouce)....

I want to know wht does it implies.....in general i hve few questions listed below after the log..::

BEGIN BW_BTE_CALL_BW204020_E 13.514

END BW_BTE_CALL_BW204020_E 13.514

BEGIN EXIT_SAPLRSAP_002 13.514

Datasource:0MATERIAL_ATTR, In: 13.514 , Out: 13.514

END EXIT_SAPLRSAP_002 13.514

Asynchronous send of data package 1 in task 0002 (1 parallel tasks)

IDOC: Info IDoc 2, IDoc No. 442405, Duration 00:00:00

IDoc: Start = 26.06.2007 19:34:06, End = 26.06.2007 19:34:06

tRFC: Data Package = 1, TID = 0A01A001012F4681A2A7362D, Duration = 00:0

tRFC: Start = 26.06.2007 19:35:03, End = 26.06.2007 19:35:04

BEGIN BW_BTE_CALL_BW204020_E 0

END BW_BTE_CALL_BW204020_E 0

BEGIN EXIT_SAPLRSAP_002 0

Datasource:0MATERIAL_ATTR, In: 0 , Out: 0

END EXIT_SAPLRSAP_002 0

Asynchronous transmission of info IDoc 3 in task 0003 (0 parallel tasks

IDOC: Info IDoc 3, IDoc No. 442406, Duration 00:00:00

IDoc: Start = 26.06.2007 19:35:04, End = 26.06.2007 19:35:04

Synchronized transmission of info IDoc 4 (0 parallel tasks)

IDOC: Info IDoc 4, IDoc No. 442407, Duration 00:00:00

IDoc: Start = 26.06.2007 19:35:04, End = 26.06.2007 19:35:04

Job finished

I want to confrm my understanding abt extraction process when infopackage gets triggerred...

Infopackge gets triggered....->R/3 job gets created/triggered....-> Info iDoc regarding request sent to BW..-> Application document(??)) posted..(( posted to what???.... passed to Trfc??....and then to psa which is seen in monitor??))...

Like wise after successfull passing of datapackage to BW ( not posting to pSA) , next datapackage is sent..... and so on...

What is TRfc?? .as far I know its function modules created to pass data to bW thru RFC call...(??)...

IF you can give me clear picture of whts happening behind the scenes, it would be great.. or helpful links would do....it would be help me to enjoy wht you ae saying and how to tune more...

THanks for your patience and help...

Already assigned points to you..

PS: I hve turned OFF buffering for master data records greater than 100, 000 and saved approx 25% for each master data records....

Regards

Gaurav

[Former Member]

Hi Nagesh,

Any comments??

(( dont want to disturb you - but i need to clear my understaning.. and many thanks for helping)

Gaurav

[Former Member]

I'm sorry my dear friend.... i forgot about this thread....

1. You need to check the 0mat_plant extractor job only as we are discussing about that. Any way, if the job compltes quickly... no need to check.

2. Some times TRFCs will be locked, we can view those in SM58. While loading, if you feel load is struct or TRFCs locked. you need to check SM58(only when you feel load is struck).

3. we have 2 transfer methods... 1. Idoc and 2. TRFC. TRFC will improve the performence of loading compare to IDOC. If you use PSA Transfer Method, it will TRFCs. We used select transfer method while creating Tranfer Rules. Please check in Forum if you have any Question on Transfer Method.

What ever you are said perfect right on data load process and TRFCs.

Note: SM58 won't give logs for Locked TRFCs. Some period Time it would be locked... you need to release that while if you feel load struck. If you check now, you won't get thing out of it. Transfer Methos PSA for all DS in BI 7.0, execpt Hierarchies( still using Idoc).

Nagesh Ganisetti.

• Assign points if it helps.

[Former Member]

Hi Nagesh,

Thanks again..

1) So TRFC are functional modules which pass data to BW( i m clear about transfer methods but dnt knw wht exactly is TRFC - like in other method we use ALE tech and use data idocs for that...) so in TRFC no data idocs are used but just TRFC...

2) I have seen when load is truck, we need to drive them (F6) for records which are recorded but not executing...

but still wonderng wht you mean by "locked"...did you mean same thing?

Thanks

Gaurav

PS: I have assigned points for your effort....