ETL workers themselves are typically ephemeral, plumbing batches between remote storage systems like Postgres, S3, and Hive. You might use local disk as scratch

ETL workers themselves are typically ephemeral, plumbing batches between remote storage systems like Postgres, S3, and Hive. You might use local disk as scratch space during the batch, but not as a sink.