Any best practices when it comes to rapid mart testing?
We have installed a few rapid marts and ran the reports but the numbers don’t match to what they have in the source. Rapid marts have a lot of objects and trying to translate the SQL’s is a lot of work. It might be work we need to do, but what’s some of the testing strategies that worked for you?
Thanks.
dellster (BOB member since 2008-10-14)
Hi, there is a new sub-forum dedicated to Rapid Marts so I’m moving your question there for further discussion, thanks.
Dave Rathbun 
(BOB member since 2002-06-06)
There is not much in the RMs by way of Business Logic - the extractions are basically 1:1. Best way I have found to test is to be as granular as possible.
What is not matching up? Is it master/dimension or transaction/fact data?
What are you matching the RM reports to in SAP? Table data as by way of se16/sqvi or to a specific transaction?
SalH 
(BOB member since 2010-11-09)
The RapidMart contains measures the same measures are used in SAP standard reports. It is just hard sometimes to find them together.
So what I do is running a report in the RapidMart that should produce the same number as a standard report in SAP, e.g. there is the cost sheet for one period in SAP, I create a similar report via the RapidMart.
When I get the same numbers, fine, if they are different I need to figure out why and often will learn that additional filters are required or the report is meant to answer an entirely different question. For testing it is important that you pick a report in SAP where you can go deeper and deeper potentially down to the individual bookings. This way you can figure out if the measures are different always or if there is a single type of bookings that are not covered.
Example 1: I compared the Sales RapidMart Revenue with the SAP Revenue numbers and they were all different. Why? Because the Sales RapidMart does report on the orders/billings/deliveries but the customer was showing me the finance numbers, where payment terms and other things are included.
Example 2: We compared Sales with Sales and for one type of sales orders we got wrong numbers, the service orders (maintenance contracts). So we had to adjust the report to get the desired result. Btw, the desired result was, as we found out after analysis of all, neither the SAP number nor the initial RapidMart report but something between.
Werner Daehn 
(BOB member since 2004-12-17)