Github user fjh100456 commented on the issue: https://github.com/apache/spark/pull/19218 @gatorsmile I'd tested the performance of 'uncompressed', 'snappy', 'gzip' compression algorithm for parquet, the input data volume is 22MB, 220MB, 1100MB, respectively run 10 times, finally 'snappy' in several cases are more excellent. The test results are as follows:(TimeUnit: ms) ![default](https://user-images.githubusercontent.com/26785576/33362659-74c0cf06-d518-11e7-9907-2f353ffed37d.png)
--- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org