Why the sum of data parts size / rows is higher than the size of the whole InfoProvider?
If DB statistics aren't recalculated on a regular basis, it is possible that the size of an InfoProvider (consisting of database tables) in DB statistics doesn't match with the real size. Data distribution analysis obtains the size of data partitions from database tables directly by a select statement. However, it also obtains the total InfoProvider size from DB statistics, which aren't always current. Therefore, the total InfoProvider size in the Data distribution analysis can be smaller than sum of its data parts.