GitHub user xuchuanyin opened a pull request:

    https://github.com/apache/carbondata/pull/2133

    [CARBONDATA-2304][Compaction] Prefetch rowbatch during compaction

    Add a configuration to enable prefetch during compaction.
    
    During compaction, carbondata will query on the segments and retrieve a 
row, then it will sort the rows and produce the final carbondata file.
    
    Currently we find the poor performance in retrieving the rows, so adding 
prefetch for the rows will surely improve the compaction performance.
    
    In my local tests, compacting 4 segments each with 100 thousand rows costs 
30s with prefetch and 50s without prefetch.
    
    In my tests in a larger cluster, compacting 6 segments each with 18GB raw 
data costs 45min with prefetch and 57min without prefetch.
    
    Be sure to do all of the following checklist to help us incorporate 
    your contribution quickly and easily:
    
     - [x] Any interfaces changed?
     `NO`
     - [x] Any backward compatibility impacted?
     `NO`
     - [x] Document update required?
    `Add a configuration, will update it later`
     - [x] Testing done
            Please provide details on 
            - Whether new unit test cases have been added or why no new tests 
are required?
    `Yes`
            - How it is tested? Please attach test report.
    `Tested in local and a 3-node cluster`
            - Is it a performance related change? Please attach the performance 
test report.
    `Compaction performance has been enhanced by 25+%`
            - Any additional information to help reviewers in testing this 
change.
           
     - [x] For large changes, please consider breaking it into sub-tasks under 
an umbrella JIRA. 
    `Not related`


You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/xuchuanyin/carbondata 0402_compaction_prefetch

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/carbondata/pull/2133.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #2133
    
----
commit 41869effa326052b46088f68dd1d6ccc5f7525e5
Author: xuchuanyin <xuchuanyin@...>
Date:   2018-04-02T12:38:17Z

    Prefetch rowbatch during compaction
    
    Add a configuration to enable prefetch during compaction.

----


---

Reply via email to