Report Portal

I got an explanation from Akshai Mirchan

I got an explanation from Akshai Mirchandani (MSFT SSAS Team) about this behaviour: ProcessUpdate copies the old data and then conceptually refreshes it based on the new dimension data. It has to do that so that it maintains the DataIDs (partitions would have to be reprocessed if DataIDs change). ProcessFull is often able to find duplicates – but ProcessUpdate no longer can find duplicates because very often they’re already present from the previous processing. Even ProcessFull doesn’t always find duplicates – or it finds false positives – e.g. using ByTable, or a multi-pass algorithm would cause it to lose duplicate knowledge. It’s a misconception that DuplicateKey guarantees detecting attribute relationship correctness – it’s just an easy builtin capability that can help you find common problems.

 

2007-2015 VidasSoft Systems Inc.