Thanks all for the help. I went through the docs and links on LO extraction and they are really useful. I understood concept, but at the same time have some queries also.
Regarding set up table, I understood that it is used at the time of init and full load. Also it acts as a layer between application table and extractor, but I did not understood the real need of this. I mean it is stroing the same data from application table redundantly. There must be some strong reason behind it, which I am missing. Can some one guide me here.
I understood the concept, but what I want to know is the advatange of extracting from setup table instead of application table directly.
Amit Saxena replied
Here is whole about Setup Tables:
Also , see these previous posts.....
The purpose of Setup Tables is to facilitate extraction speed. Rather than getting the data directly from the source tables (MARA, EKKO, EKPO, MARC, VBRP, VBRK, MSEG etc.), the data is conveniently stored in SETUP TABLES. Which you can use from time to time when needed.
Just imagine this (no setup tables involved), you ran INIT Request. The program extracts data from source tables directly. Suddenly, an error occured. You need to do REINIT again and run thru those large source tables again.
The setup tables are filled with data used for a full update to BW. Normally you use a user free time to run a setup for all historic data. After setup you schedule a initilization of delta (with or without data transfer). After that all transaction data is posted to the delta queues in the source system. After a sucessful full upload you can delete the setup tables. They only use disc space. If you once will do another full load, you can easily start another run to fill the setup tables without any effect to the delta mechanism.
These Set up tables contains fields: CREATED BY, DATE CREATED, EDIT BY, EDIT DATE.. if all the tables contains these fields then delta will be very easier without the setup tables...
LO set-up tables can be deleted after data validation in BW. They are cluster tables storing history in an optimal manner. Finally, if your R/3 and BW have simultaneous go-lives, you do not need the set-up table.
Using the transaction NPRT(Log for setup of statistical data)in R/3 side, you can find all the log infomation regarding setup tables and their selections.
You can delete the contents of the setup tables anytime you want with no impact on R/3. However, when you go to load the setup tables, make sure that none is entering/changing/deleting deliveries. When you load the setup tables (tx OLI8BW or drill down in SBIW), the only parameters you need to fill in are in the "Control of the setup run" section.
Name of run (anything you want)
New run (check)
Date of termination (put tomorrow's date)
Leave the rest of the parameters alone, and then execute (Hint: at the top select Program and Execute in batch (F9))
This will submit a job to fill the contents of the setup table for deliveries.
When this is finished, execute the InfoPackage in BW for Initializing the Delta (with data). This infopackage will pull the contents of the setup table into your cube and setup the delta queue in R/3 for subsequent delta loads.
Hope it helps!!!!