Data engineers should be building pipelines and delivering business value, not fidgeting with some JVM or Spark parameter that saves them runtime on a join (or

Data engineers should be building pipelines and delivering business value, not fidgeting with some JVM or Spark parameter that saves them runtime on a join (or for that matter, from what I've seen at a certain bigco, building their own custom join algorithms). That's why I said it's only economical for big companies to run efficient abstractions and everyone else just throws more compute at the issue.