I spent 6 hours studying PySpark join strategies. Here's what I learned

From broadcast to shuffle-hash: understanding how Spark executes joins under the hood