Optimizing PySpark Code

In this session we cover ways to optimize PySpark code. This includes descriptions of situations where slowness may occur, for example, uneven partitions and skewed joins. To combat these issues I explain repartitioning/coalescing and broadcast joins. I also explain how to place your data in memory or on disk to cache commonly used data sets. Finally, I show the interface where you can monitor memory and CPU usage to make sure you are using the optimal cluster size.

Lastly, I show how to use multiple languages inside of one Databricks notebooks including SQL and R code.