[jira] [Commented] (HIVE-396) Hive performance benchmarks

["dhruv kapatel (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HIVE-396?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15162795#comment-15162795 ] 

dhruv kapatel commented on HIVE-396:
------------------------------------

Here they have considered real time for benchmark. I think real time you get from time command may affected by other processes also. Does it fair comparison by considering real time?

in time command user+sys represents total cpu time right. Then it should be same as total cpu time spent in job counter. But i am getting different results. 
I've mentioned more details in this question.
http://stackoverflow.com/questions/35500987/which-one-should-i-use-for-benchmark-tasks-in-hadoop-usersys-time-or-total-cpu   



> Hive performance benchmarks
> ---------------------------
>
>                 Key: HIVE-396
>                 URL: https://issues.apache.org/jira/browse/HIVE-396
>             Project: Hive
>          Issue Type: New Feature
>            Reporter: Zheng Shao
>            Assignee: Yuntao Jia
>         Attachments: AlansMRcode.tgz, hive_benchmark_2009-06-18.pdf, hive_benchmark_2009-06-18.tar.gz, hive_benchmark_2009-07-12.pdf, hive_benchmark_2009-07-21.tar.gz
>
>
> We need some performance benchmark to measure and track the performance improvements of Hive.
> Some references:
> PIG performance benchmarks PIG-200
> PigMix: http://wiki.apache.org/pig/PigMix



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)