You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@avro.apache.org by "Doug Cutting (JIRA)" <ji...@apache.org> on 2014/11/26 20:20:12 UTC

[jira] [Commented] (AVRO-1611) Avro-mapred should provide CombineAvroKeyInputFormat

    [ https://issues.apache.org/jira/browse/AVRO-1611?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14226678#comment-14226678 ] 

Doug Cutting commented on AVRO-1611:
------------------------------------

Sounds reasonable.  Can you please provide unit tests for this?  Thanks!

> Avro-mapred should provide CombineAvroKeyInputFormat
> ----------------------------------------------------
>
>                 Key: AVRO-1611
>                 URL: https://issues.apache.org/jira/browse/AVRO-1611
>             Project: Avro
>          Issue Type: Improvement
>            Reporter: Bartłomiej Romański
>            Priority: Minor
>         Attachments: CombineAvroKeyInputFormat.java
>
>
> A serious issue with Hadoop is dealing with a huge number of small files (they slow down processing, overload namenode etc...). 
> A common remedy for this is to use CombineFileInputFormat. However, this is an abstract class to need to be concretized for each InputFormat it is wrapping. I believe Avro should provide CombineAvroKeyInputFormat like Hadoop is providing CombineSequenceFileInputFormat and CombineTextInputFormat.
> I've attached a basic implementation.



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