Hi,

During initial implementation there was a decision to pseudonymize column USNAM in multiple tables. Currently it was agreed that we need to have actual (not pseudonymized) users data in Celonis just like it is in SAP.

Usually unticking the Pseudonymize box in extraction setup would do the trick and I will do it for some smaller tables for which we have full loads. However we have couple tables for which we are doing only delta loads, as they are way to big for full loads (over 40 GB). Doing full load for them would take ages and I'm not entirely sure how Celonis would manage with such a huge load done at once.

The idea is to create one time mapping by downloading second iteration of that big table (BKPF is original one, the second one lets call it BKPF_RAW), that would only contain the 4 key columns and USNAM that is not pseudonymized. Next step would be to add new column (lets call it USNAM_RAW) to original BKPF, then update it with data from BKPF_RAW.USNAM by matching the two tables by unique keys. Next step would be deleting the original BKPF.USNAM and then renaming the BKPF.USNAM_RAW to USNAM in order to keep the name for all the transformations that we have. And also the last part would be to uncheck the Pseudonymize box in delta extractions, so that all the new data is consistent with the data to which we've changed or USNAM column.

Have any one tried this approach? It seems feasible at least to the point of replacing the data, but I'm wondering about any potential risk with data synchronization or any other issues that may occur later in transformations due to all of the shenanigans (keeping the same data type for BKPF.USNAM and BKPF.USNAM_RAW might be important). I'm already preparing data for testing in QA env but was wondering about your input.

Thanks,

Karol