The first 1 million GBIF records took only about a minute to process. However, as the indexes grew, each successive million records slowed down. I’m now planning to do them in smaller batches.