I recently was tasked with performing an ETL task that should be done as efficiently and quickly as possible. The work led me to learn more about parallel and distributed processing in Clojure. In addition to having a greater appreciation for what Clojure enables (once again), I also pushed the boundaries of what I thought is possible using the available tools. I ultimately ended up writing a Spark job whose executors are each running N threads (currently, N=3). But the path to that solution taught as much by what didn’t work as much as what did work.