Hey, you're right. Thanks for letting me know that I need to bring more clarity to that part. So, memory or RAM (Deserialised) <--> Disk Spill (Serialised into temporary spill files) I've updated the article on this, so check it out: https://vaishnave.page/sparks-fly#heading-serialization-between-memory-and-disk

Hi Kolisetty Sasiram The shuffling does happen in the case of sort-merge join as well. I didn't include it because only if the data is not already partitioned by the join key, Spark performs a shuffle operation to redistribute the data across the cluster based on the join key. But, since I'm listing out the process, it makes more sense to add it. Thank you for your input!