Long Time : Start Data Selection for Header Tables

[Former Member]

Good Morning,

I am running the Activity "Start Data Selection for Header Tables" since the March 29 and not yet complete.

I look at the SM50 is reading sequentially EQUI tables, EQUZ and EQKT (EQUI table has more than 200 million records).

SELECT

FROM

"equi"

WHERE

"mandt" = ? AND "equnr" = ?

I see the log in sm37 following:

04.04.2010 10:48:00 78,200,000 records read

04.04.2010 10:57:17 78,300,000 records read

04.04.2010 11:06:58 78,400,000 records read

04.04.2010 11:17:01 78,500,000 records read

04.04.2010 11:29:42 78,600,000 records read

04.04.2010 11:41:20 78,700,000 records read

I have a applied to the Support Package 13.

I have to adjust some settings? Any ideas?

Thanks,

Enrique.

[former_member126060 · Accepted solution]

Hi Hugo,

First things first - Normally EQUI tables have huge number of entries and its selection is purely dependent on some other tables and the selected date.

Now, 5 days is slightly enough for the data selection activities so i doubt is the activity still in process ?

check if you found any bottleneck ? whats the status of load in the sender system ?

Check SAPnote # 916763 - TDMS performance u201Ccomposite SAP noteu201D. Specially Data selection part. this is one of the best note i found, atleast it have the things expressed clearly with reasons. JUst check and take the actions accordingly.

Best Regards

Niraj

[Former Member · Accepted solution]

Hi Niraj

We have reviewed the letter and the suggestions and have implemented. However, the process is still ongoing.

We note that read on the table EQUI very slow.

So now is reading the amount is 128 million records.

Do not know what may be happening.

Thank you.

Enrique.

[former_member126060 · Accepted solution]

Hello Hugo,

Did you used P_FULL = X during fill header of the EQUI ??

Can you try to create secondary index on MANDT and EQUNR on tables EQUI ,EQUZ and EQKT and check if it can help.

Best Regards

Niraj

[Former Member · Accepted solution]

Hi Niraj,

Yes, I used P_FULL = X during fill header of the EQUI.

The 3 tables are as the primary key fields: MANDT and EQUNR.

I look in the log of SM37 currently has read 128,700,000 records.

I show you the last 8 lines:

09.04.2010 08:24:04 128,000,000 records read

09.04.2010 08:32:57 128,100,000 records read

09.04.2010 08:47:39 128,200,000 records read

09.04.2010 09:06:25 128,300,000 records read

09.04.2010 09:26:03 128,400,000 records read

09.04.2010 09:45:58 128,500,000 records read

09.04.2010 10:04:03 128,600,000 records read

09.04.2010 10:20:34 128,700,000 records read

and is recorded in Table DMC_INDXCL the amount of 435.958 registrations.

For some reason the process is very slow.

Thanks,

Enrique.

[bogdan_brebenel]

I experienced the same long runtimes for large tables.

If secondary indexes do not help, I suggest opening an OSS Message for SAP in regard of this problem.

Edited by: Bogdan Brebenel on Apr 9, 2010 9:43 PM

[Former Member]

Hi,

Yes, I have opened an OSS message.

Thanks,

Enrique.

[former_member126060]

Hi Enrique,

For me things look fine now from TDMS view :

1. P_FULL is set to X

2. INdexes are present MANDT and EQUNR.

3. You don't have problem with the job or WP.

I'll advise to ask your DB team to get in action and check the things from the DB perspective.

Best Regards

Niraj