Time for Data Extraction

[Former Member]

Hello,

i have a problem/question at the data load into an InfoCube. The data load contains two requests.

1. approx. 8.500.000 data entries

2. approx. 3.500.000 data entries

Extraction time of the 1st request takes approx. 7 hours. Second request takes 15 hours.

Data of both are requests are homogenous. Knows any a reason why the extraction of the 2nd request takes 2 times longer, altough the amount of records is less than the half.

Thanks for advise.

[Former Member]

1. package size is in both cases the same

2. extraction of the requests is sequential, table(s) of the source system are not locked

3. no routines in use which can corrupt performance at data loading

4. source system is not the bottleneck, because the performance decrease appears at the update/insert into fact and dimension table(s)

5. PSA-Settings: PSA and Then into Data Target/InfoObject (by Package)

Message was edited by:

Markus Linortner

[Former Member]

Normally we can't proportionally say, since it tool this much time in the first request it should take this much for this number of records. But in this example, there is a big difference in the load time. In addition to what has been mentioned.

Sometimes, once the job it triggered, in the source system depending on the number of available background process, the job will be in just release status, and wait until there is available process. So basically, the time the job triggered is also a factor on the number of hours the load will take.

thanks.

Wond

[Former Member]

Adding to what Satish said, how did you define the process in the Infopackage? Only PSA? PSA and Data Targets in parallel?, etc... Normally, the fastest one is Only PSA and then Data Targets (3rd one counting from the top)...

Also, check with your Basis guys the data packet size define for this load... Check SAP Note 417307.

Good luck.

Luis

[Former Member]

It depends on many factor.. Few of them are

1. The source system load is high during the second run

2. Source table you are trying to read is locked during the second run

3. The are some routines in the Transformation which has some records which is causing performance..

Check in the monitor screen where the time is consumed. Go to the source system and analyse the time in the job log for both the load. This will give u some hint.

Thanks,

Satish