wangyum commented on issue #24672: [SPARK-27801] Improve performance of InMemoryFileIndex.listLeafFiles for HDFS directories with many files URL: https://github.com/apache/spark/pull/24672#issuecomment-495052170 Thank you @rrusso2007 @JoshRosen I did simple benchmark in our production environment(Hadoop version is 2.7.1): ``` 19/05/22 19:53:18 WARN InMemoryFileIndex: Elements: 10. default time token: 41, SPARK-27801 time token: 18, SPARK-27807 time token: 30 19/05/22 19:53:29 WARN InMemoryFileIndex: Elements: 20. default time token: 22, SPARK-27801 time token: 10, SPARK-27807 time token: 24 19/05/22 19:53:30 WARN InMemoryFileIndex: Elements: 50. default time token: 47, SPARK-27801 time token: 13, SPARK-27807 time token: 25 19/05/22 19:53:33 WARN InMemoryFileIndex: Elements: 100. default time token: 54, SPARK-27801 time token: 10, SPARK-27807 time token: 30 19/05/22 19:53:42 WARN InMemoryFileIndex: Elements: 200. default time token: 86, SPARK-27801 time token: 19, SPARK-27807 time token: 40 19/05/22 19:53:52 WARN InMemoryFileIndex: Elements: 500. default time token: 254, SPARK-27801 time token: 30, SPARK-27807 time token: 90 19/05/22 19:54:06 WARN InMemoryFileIndex: Elements: 1000. default time token: 507, SPARK-27801 time token: 165, SPARK-27807 time token: 117 19/05/22 19:54:18 WARN InMemoryFileIndex: Elements: 2000. default time token: 1193, SPARK-27801 time token: 114, SPARK-27807 time token: 216 19/05/22 19:54:34 WARN InMemoryFileIndex: Elements: 5000. default time token: 2401, SPARK-27801 time token: 430, SPARK-27807 time token: 565 19/05/22 19:54:56 WARN InMemoryFileIndex: Elements: 10000. default time token: 4831, SPARK-27801 time token: 646, SPARK-27807 time token: 1202 19/05/22 19:55:40 WARN InMemoryFileIndex: Elements: 20000. default time token: 9121, SPARK-27801 time token: 1535, SPARK-27807 time token: 1920 19/05/22 19:56:45 WARN InMemoryFileIndex: Elements: 40000. default time token: 18873, SPARK-27801 time token: 2784, SPARK-27807 time token: 3997 19/05/22 19:58:18 WARN InMemoryFileIndex: Elements: 80000. default time token: 33658, SPARK-27801 time token: 6476, SPARK-27807 time token: 8326 ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
