Do Not Use Datamart Normalization 'High'

Why

While it is possible to set the normalization of a Datamart to High, this should not be done. Doing so can drastically slow down the performance of Datamart queries, especially when processing large (i.e. 1M+ rows).

The reason is quite simple. A Datamart with normalization set to High will not actually contain the data from its underlying Data Source. Hence every query needs to join the Datamart with the underlying Data Source. And this gets more and more costly as the number of rows increases.

What to Use

It is strongly recommended to prefer the Medium normalization level.

Procedures for Changing the Datamart Normalization

After changing the normalization level for a Datamart, you need to deploy the Datamart, then run the refresh job.