Question:
I am submitting this again after having created a login account
Our organsiation is attemtping to determine the best method of running R3 to BW extracts (and associated processing, updates etc) when we operate a single R3 and BW instance over a single region with many time zones.
Up to now, we have pursued the model of a single nightly extract and update process. However, we need to wait until the far western country in the time zone has finished work for the day and end of day processing has completed, prior to being able to initiate the BW extracts. This however means that by this time, the far eastern country in the time zone is almost ready to start business the following day, and we don't have enough time to run the porcesses, update cubes and reports before that country starts.
To get around this, we have considered the following optoins:
1. Run the extract and update processes many times over the night. However, this means that the data in the cubes could change each time. Also, it may not be the most efficient method.
2. CHange the extract process to fileter only on Coutry code, and update the ODS and cubes with that filtered extracted data. We understand however that this is a problem with the R3 delta queue in that you can't filter delta queue data and have the rejected data from the filter process retained on the queue for subsequent extracts.
3. Run the extract (from option 1) many times, but only update the ODS, but not eh cubes. Schedule the cubes to be updated once and only when all extracts and processes have completed.
are there any others that you can think of? What are the benefits and disadvantages of each?
Answer:
Hello Sharpy,
i'm currently on a project with a similar issue and we went for solution 3. Certainly if you use Process Chains, you can be very flexible on doing the loads to the Cube inly in the last "load-flow"...