Spark Join Strategies: The Query Is Innocent. The Cluster Is Not.

1. First, stop thinking of JOIN as a single operation Consider a familiar data engineering workload. We have billions of transactions in a fact table and a relatively small token metadata table contai