Its ok, but seems like its getting covered like a common theory session skipping some parts here and there. I feel it could have been better in case it would have been demonstrated practically. I don't see anything talked about the performance improvement each of these optimizations caused, neither was skewness or spill shown in spark UI.