[jira] Updated: (HADOOP-485) allow a different comparator for grouping keys in calls to reduce

["Tahir Hashmi (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HADOOP-485?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Tahir Hashmi updated HADOOP-485:
--------------------------------

    Attachment: Hadoop-485-pre.patch

Here's a proof-of-concept patch for this requirement. The intent is to get the approach reviewed. What's being done here is that instead of just comparing keys, we're now also comparing values if keys turn out to be equal. This is done both while doing sortAndSpillToDisk() in MapTask.java and while doing a merge in SequenceFile.java.

The trick is to alter the comparison functions in BasicTypeSorterBase.compare() and SequenceFile.MergeQueue.lessThan() to account for values as well. The attached patch has the code changes done for this.

> allow a different comparator for grouping keys in calls to reduce
> -----------------------------------------------------------------
>
>                 Key: HADOOP-485
>                 URL: https://issues.apache.org/jira/browse/HADOOP-485
>             Project: Hadoop
>          Issue Type: New Feature
>          Components: mapred
>    Affects Versions: 0.5.0
>            Reporter: Owen O'Malley
>         Assigned To: Tahir Hashmi
>         Attachments: Hadoop-485-pre.patch
>
>
> Some algorithms require that the values to the reduce be sorted in a particular order, but extending the key with the additional fields causes  them to be handled by different calls to reduce. (The user then collects the values until they detect a "real" key change and then processes them.)
> It would be much easier if the framework let you define a second comparator that did the grouping of values for reduces. So your reduce inputs look like:
> A1, V1
> A2, V2
> A3, V3
> B1, V4
> B2, V5
> instead of getting calls to reduce that look like:
> reduce(A1, {V1}); reduce(A2, {V2}); reduce(A3, {V3}); reduce(B1, {V4}); reduce(B2, {V5});
> you could define the grouping comparator to just compare the letters and end up with:
> reduce(A1, {V1,V2,V3}); reduce(B1, {V4,V5});
> which is the desired outcome. Note that this assumes that the "extra" part of the key is just for sorting because the reduce will only see the first representative of each equivalence class.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.