are foreign keys etc in the model obstacles in the vertipac engine's aggregations

[db042190]

hi, it seems getting around using very dense keys (eg customer id from sales order to customer dimension) is impossible when modeling...unless we flatten the data model so its no longer a star.  But as far as i know, dense non repeating info in the vertipac engine is counter productive to all the dictionary algorithms that make aggregations so quick.  But i'm also reading that the star is best practice.   Are we shooting ourselves in the foot when using a star schema in our models?  Or is the engine smart enough to understand that the keys are there for relationships only?