PowerPivot Data Modelling for Performance

PowerPivot allows us more flexibility than SSAS UDM when it comes to data modelling. We are no longer constrained to dimensions and facts. As Thomas Ivarsson showed us in his PowerPivot Schema Flexibility post, we can avoid certain data manipulations when building a model in PowerPivot. The same holds true for relational database scenarios, as well. We can write SQL queries and analyse data without complying with any particular model. However, we have the concepts of a data mart and a star schema, which are not based on SSAS UDM but rather the other way around. In this post I will compare a normalised and a denormalised model in PowerPivot and reason around the relevance of data modelling and denormalisation with Microsoft in-memory technologies.