Data extraction & source system performance

[Former Member]

Dear Experts,

I have a ECC as source system in APO BW

And there is this hypothesis .

"Frequent scheduling of data loads from infosources to infocubes MAY put load on ECC system".

e.g. I have a datasource for extracting sales data. I decide to extract sales data everyday. Lets say I use a datasource readily delivered by SAP.

- Does this cause performance issues in ECC ?

- Is this really the case that scheduling of data load to data targets in BW makes queries on SD tables and hence can cause unwanted locking of data?

- What is the right way of explaining this phenomenon to someone who has never heard of APO and BW ?

Thanks

Guru

[rajkj · Accepted solution]

Hi Guru,

Your hypothesis is correct.

Any data request originated from BW system involves with execution of few tasks in the data source system (for instance ECC). Typically execution calls involve data pull from different source tables and populate the extraction structure (i.e. data extraction and accumulation) and qRFC or tRFC calls (i.e. prepare data for transfer).

Probably, your Basis and BW administrator teams can collect the execution times and data volumes by activating execution trace. It will give you the required evidence to prove your hypothesis.

Sales data extraction structure will be updated in real time and BW system pulls the data during off peak hours once a day.

Thanks,
Rajesh

[Former Member · Accepted solution]

Thanks Rajesh,

Many thanks for your response. Clear to me now.

I was actually mixing SQL query time and data extraction (ETL) time here.

The phrase "datasource" made it sound (in my mind) as some "advanced and optimized" way of reading data from the source tables and hence I need not be bothered about run times and performance and even if there is some run time in data extraction using datasource-infosource technology, it is not of the order that happens when a table with million entries is queried using SQL or a report program. This was the context  of the question. Though such is not written anywhere on help.com is so many words.

Could you once again elaborate a bit on what does this mean.

Sales data extraction structure will be updated in real time

Just wanted to be sure I am not completely ignorant of something important I need to keep in mind.

My understanding is when I click on "start schedule" in infopackage, the data will be extracted from the source i.e.all the way from ECC source tables and then to infosource in BW and then updated to the target cubes (if required). I can schedule this extraction and loading of data in off peak hours using process chains.

Once again many thanks. Really appreciate your invaluable contributions on this forum. I learnt more SAP reading responses like yours. 

Regards

Guru

[rajkj · Accepted solution]

Hi Guru,

First of all, thanks for the nice comments.

Regarding the statement sales data extraction structure will be populated in real time, I observed a queue entry creation as soon as a sales order was saved. This queue entry helps to update extraction structure and can also push the updated data up to BW PSA based on the configuration.

Then, another important feature that impact the source system performance is the presence of transformation tasks (such as conversions, aggregations, mappings, etc)  on extracted data before its transfer from source system to destination.

Regards,
Rajesh

[Former Member · Accepted solution]

Thanks Rajesh

Now crystal clear to me.

Thanks

Guru