Grokbase Groups Hive user June 2009
FAQ
Hi all,

Yuntao Jia, our intern this summer, did a simple performance benchmark for Hadoop, Hive and Pig based on the queries in the SIGMOD 2009 paper: A Comparison of Approaches to Large-Scale Data Analysis

The report and the performance test kit are both attached here:
http://issues.apache.org/jira/browse/HIVE-396


We tried our best to get good performance out of Hive and Pig, and we keep the hadoop program as close as it is from the SIGMOD paper. We welcome all suggestions on how we can improve the performance more by both changing the configuration or improving the code.


While we tried our best to be fair, system settings and environments do affect the result a lot. So we encourage everybody to try out the performance test kit on their own cluster, and we will appreciate if everybody can share their results.


Here is the summary. The details are in the report hive_benchmark_2009-06-18.pdf from the link above.

Query: GREP SELECT
Hadoop: 136.1s
Hive: 125.4s
Pig: 247.8s

Query: RANKINGS SELECT
Hadoop: 26.1s
Hive: 31.0s
Pig: 38.4s

Query: USERVISITS AGGREGATION
Hadoop: 533.8s
Hive: 768.8s
Pig: 855.4s

Query: RANKINGS USERVISITS JOIN
Hadoop: 470.0s
Hive: 471.3s
Pig: 763.9s

Please take a look at hive_benchmark_2009-06-18.pdf from the link above for details. Let's keep discussions on http://issues.apache.org/jira/browse/HIVE-396 so it's easier to keep track.


Zheng

Search Discussions

  • Owen O'Malley at Jun 19, 2009 at 5:30 pm
    This bounced since I wasn't subscribed. It should have been moderated through...

    -- Owen


    ---------- Forwarded message ----------
    From: Owen O'Malley <owen.omalley@gmail.com>
    Date: Fri, Jun 19, 2009 at 10:03 AM
    Subject: Re: A simple performance benchmark for Hadoop, Hive and Pig
    To: core-user@hadoop.apache.org, pig-user@hadoop.apache.org,
    hive-user@hadoop.apache.org

    On Thu, Jun 18, 2009 at 9:29 PM, Zheng Shao wrote:


    Yuntao Jia, our intern this summer, did a simple performance benchmark for Hadoop, Hive and Pig based on the queries in the SIGMOD 2009 paper: A Comparison of Approaches to Large-Scale Data Analysis
    It should be noted that no one on the Pig team was involved in setting
    up the benchmarks and the queries don't follow the Pig cookbook
    suggestions for writing efficient queries, so these results should be
    considered *extremely* preliminary. Furthermore, I can't see any way
    that Hive should be able to beat raw map/reduce, since Hive uses
    map/reduce to run the job.

    In the future, it would be better to involve the respective
    communities (mapreduce-dev and pig-dev) far before pushing benchmark
    results out to the user lists. The Hadoop project, which includes all
    three subprojects, needs to be a cooperative community that is trying
    to build the best software we can. Getting benchmark numbers is good,
    but are better done in a collaborative manner.

    -- Owen
  • Ashish Thusoo at Jun 19, 2009 at 6:06 pm
    This numbers are definitely preliminary and the reason that we send them out was to involve the community from the get go and have them critique this work. The mistake though was sending this out on the users list as opposed to the dev lists.
    Regarding better than map/reduce I think that the number is better than thae particular way the query was implemented in the SIGMOD paper. It is more of a reflection of the implementation there as opposed to map/reduce in general.

    In keeping with Owen's comments we should move this discussion to the dev lists, users is not an appropriate forum for it.

    Ashish

    ----- Original Message -----
    From: Owen O'Malley <owen.omalley@gmail.com>
    To: core-user@hadoop.apache.org <core-user@hadoop.apache.org>; pig-user@hadoop.apache.org <pig-user@hadoop.apache.org>; hive-user@hadoop.apache.org <hive-user@hadoop.apache.org>
    Sent: Fri Jun 19 10:03:06 2009
    Subject: Re: A simple performance benchmark for Hadoop, Hive and Pig

    On Thu, Jun 18, 2009 at 9:29 PM, Zheng Shao wrote:

    Yuntao Jia, our intern this summer, did a simple performance benchmark for
    Hadoop, Hive and Pig based on the queries in the SIGMOD 2009 paper: A
    Comparison of Approaches to Large-Scale Data Analysis

    It should be noted that no one on the Pig team was involved in setting up
    the benchmarks and the queries don't follow the Pig cookbook suggestions for
    writing efficient queries, so these results should be considered *extremely*
    preliminary. Furthermore, I can't see any way that Hive should be able to
    beat raw map/reduce, since Hive uses map/reduce to run the job.

    In the future, it would be better to involve the respective communities
    (mapreduce-dev and pig-dev) far before pushing benchmark results out to the
    user lists. The Hadoop project, which includes all three subprojects, needs
    to be a cooperative community that is trying to build the best software we
    can. Getting benchmark numbers is good, but are better done in a
    collaborative manner.

    -- Owen
  • Zheng Shao at Jun 19, 2009 at 6:08 pm
    I completely agree with Owen on this point. Let's move all discussions to dev lists and jira: http://issues.apache.org/jira/browse/HIVE-396

    I was confused by seeing so many automatic emails in the dev mailing list.

    Zheng
    -----Original Message-----
    From: Owen O'Malley
    Sent: Friday, June 19, 2009 10:03 AM
    To: core-user@hadoop.apache.org; pig-user@hadoop.apache.org; hive-user@hadoop.apache.org
    Subject: Re: A simple performance benchmark for Hadoop, Hive and Pig

    On Thu, Jun 18, 2009 at 9:29 PM, Zheng Shao wrote:

    Yuntao Jia, our intern this summer, did a simple performance benchmark for
    Hadoop, Hive and Pig based on the queries in the SIGMOD 2009 paper: A
    Comparison of Approaches to Large-Scale Data Analysis

    It should be noted that no one on the Pig team was involved in setting up
    the benchmarks and the queries don't follow the Pig cookbook suggestions for
    writing efficient queries, so these results should be considered *extremely*
    preliminary. Furthermore, I can't see any way that Hive should be able to
    beat raw map/reduce, since Hive uses map/reduce to run the job.

    In the future, it would be better to involve the respective communities
    (mapreduce-dev and pig-dev) far before pushing benchmark results out to the
    user lists. The Hadoop project, which includes all three subprojects, needs
    to be a cooperative community that is trying to build the best software we
    can. Getting benchmark numbers is good, but are better done in a
    collaborative manner.

    -- Owen

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupuser @
categorieshive, hadoop
postedJun 19, '09 at 4:30a
activeJun 19, '09 at 6:08p
posts4
users3
websitehive.apache.org

People

Translate

site design / logo © 2021 Grokbase