Story of SAP BW foundation
Hello BW fellows,
HANA is next generation, it will probably replace SAP BW.
To me, it is probably, not definitely. Why?
If we review the foundation of SAP BW, we will realize the brilliance and great work contributed by SAP engineers in the past.
What is SAP foundation?
I think it is LIS.
We all know BW process chain extraction delta data from ECC daily.
But do you know how the delta change be captured? timestamp it not enough.
It’s a long story,
Prior to 2000, SAP created a component called LIS (logistic information system), maybe also SIS (sales information system), this component are located inside ECC.
SAP want to supply a small reporting system based on LIS.
Inside this component, so called communication structure are created to capture all necessary field.
Very complex ABAP codes are write in the SD,MM core to capture the before and after image from table such as XVBAP when document changes happened.
That is to say, when document changed, the changes are written to LBWQ queue immediately.
This process is the starting point of current BW extraction.
It is a great idea to get the change information when change happens and save the changes into a pool.
orders, billing, inventory movement… SAP spent more than 20 years to develop various extractor to get the delta.
it’s valuable asset.
I believe even with HANA, extractors and LIS will still provide data to target system.
LIS dataSources was, for sure, a great accelerator for BI implementation projects. This includes the delta management and the pre-building of some indicators.
However, I have seen a lot of misunderstandings, fears (black box effect: how does it work ?), and misuses (adding fields to the dataSource with the BW user exit..). This could have lead to a reputation of unreliable feature I have seen among customers or consultants.
Also, the fact that you need to theorically lock the users to initialize your data flow was always a pain and sometimes a blocking point for fast moving data model.
For delta management, DB triggers are easier to implement and to understand (target = source). But then, you will have more work to do on your data model (but at least, you will understand it 😉