Phil Seamark explains a recent change:
If you’ve been working with Direct Lake in Microsoft Fabric, you’ll know its magic resides in its ability to quickly load data. It loads data into semantic models from OneLake when needed. This feature eliminates the overhead of importing. But until recently, the first query on a cold cache might feel sluggish. Why? One reason for this is that Direct Lake must build a join index. This index is added to the model during the first query. This index is a critical structure that maps relationships between tables for efficient lookups.
Earlier, this process was single-threaded and slow, especially on large tables with high cardinality. The good news? That’s changed.
Read on to see how, what a join index is, and what this impact looks like in practice.