Back in March of this year I created an article and a webinar for the Pragmatic Works Blog.
The original concepts still hold, Power BI and Excel Power Query are very powerful tools for building data models and sharing data discovery and visualizations. As you move from initial prototyping to a more mature model you may find that performance can be bogged down if you don't manage the structure of your reporting information. This talk provides Tips and Techniques to manage this information (link to updated slides).
Since the original webinar I have presented this topic at a few user groups and SQL Saturdays and was asked some very good questions.
I'd like to answer these questions here, and to update the slide deck as well.
1) In Providence at the BI User Group I was asked: What happens in Kasper de Jonge's Memory Usage Tool macro when there is a join between two queries loaded into the model?
If the join is done in the data load, the data model contains only one table, as in the example discussed in the original presentation.
If there is a relationship created between two tables loaded in to memory, the joined tables are both shown in the pivot tables that result from the macro, and how much memory taken by each column is shown as a datatype and total for the column.
2) In NYC at the Power BI User Group a member asked: Does the HierarchySlicer cause performance problems?