have been using R/3 DF from a long time, and their performance looks good to me. from DS 4.0, extractors were introduced. have gone through the pros and cons of both the methods.
we have a requirement to dump the data (all columns) from 15-20 SAP tables (mostly MARA,MARC,MARM,KNVV,KNA1…AUSP). thinking of creating views for these tables and create View based extractors.
Which is the best option. R/3 DF or the above method ? we have around 90-100K records in each table.
Don’t bother with the effort of extractors then. Even a full load takes only a few minutes in total for those tables (well maybe AUSP not).
In fact, I have no clue why those extractors should be used, unless you have a very good bug free performing adapter that does not fail on a massive amount of changes and the old fashioned way of extracting does not have a good performing delta.
I am not trying to act evil here. But we sadly realized the behaviors of extractors. They suit the requirements of many but not all. They are basically custom written ABAP Programs which they package/sell to us!
Issue 1: Full Load Extractors were good I believed. Because you get the full pull always. What could be the possible issue? One year down the line there was a requirement to refer to the Foreign Key relation to another table for which column didn’t existed in the Extractor but existed in the physcial table :(. So I had to convert my Extractor Job to an ABAP(R/3 Dataflow) Job
Issue 2: When there are failures, Extractors are pain in the neck. DS monitors Extractors via ODQMON whereas BW monitors extractors by RSA7 I believe. I maybe wrong, I forgot the T-Code. Deleting an incomplete request is something crazy. You will have to approach BASIS and I hope you will know how that is going to be!
Issue 3: If there will be a BW environment existing in your landscape, then you will have to be really careful in using extractors otherwise you are going to step on each other when using some delta extractors.
Issue 4: You need to master the Delta Mechanism. Otherwise you are going to start losing records. We are still using 2LIS_03_BF extractor which is an inventory process and replicating that using the tables is going to be a whole project by itself!
Hey said too many Cons. Let me try to say some pro!
Pro 1: Felt they are faster than ABAP. Didn’t proved it to myself!
Pro 2: Don’t need to generate the ABAP and register on the App Servers!
Even i feel ABAP program (standard way of talkingto SAP systems) is much faster than Extractors. and when i raise a ticket to SAP, they suggested R/3 over extractors for huge volumes of data.
(BOB member since 2012-03-20)
(BOB member since 2008-07-17)