Hi,

I've just testing mapred performance in mapr and I'm very surprised that it takes long time for the shuffle-phase. Our cluster has 7 nodes . The total Memory is 27 GB (4*7GB) and has 1,4 TB disk capacity (each has 200 GB). I run the standard random-writer|sort in the hadoop-example.jar with the command for

- hadoop jar $HADOOP_INSTALL/hadoop-*-examples.jar randomwriter random-data

- hadoop jar $HADOOP_INSTALL/hadoop-*-examples.jar sort random-data sorted-data

I receive following results:

-Time taken for random-data generation : 22min, 9sec

-Time taken for random-data Sort : 6hrs, 24mins, 20sec

Do you think why it takes me a long time for sorting out the data in mapr?

I've just testing mapred performance in mapr and I'm very surprised that it takes long time for the shuffle-phase. Our cluster has 7 nodes . The total Memory is 27 GB (4*7GB) and has 1,4 TB disk capacity (each has 200 GB). I run the standard random-writer|sort in the hadoop-example.jar with the command for

- hadoop jar $HADOOP_INSTALL/hadoop-*-examples.jar randomwriter random-data

- hadoop jar $HADOOP_INSTALL/hadoop-*-examples.jar sort random-data sorted-data

I receive following results:

-Time taken for random-data generation : 22min, 9sec

-Time taken for random-data Sort : 6hrs, 24mins, 20sec

Do you think why it takes me a long time for sorting out the data in mapr?

you should see a huge improvement.