You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@beam.apache.org by "Daniel Halperin (JIRA)" <ji...@apache.org> on 2016/08/22 16:45:20 UTC
[jira] [Commented] (BEAM-573) TextIO.read() should support sharding
[ https://issues.apache.org/jira/browse/BEAM-573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15431156#comment-15431156 ]
Daniel Halperin commented on BEAM-573:
--------------------------------------
I think this is working-as-intended. It is an important part of the model that the runner decides input sharding. Runners can ask input sources their size, and they *should* set the {{desiredBundleSizeBytes}} to a reasonable value if they cannot scale.
(I believe that the {{DirectRunner}} does not do this today, but that is also working as intended. It's not a scalable performance runner but rather a model-checking runner, which is also designed to not be horribly slow)
> TextIO.read() should support sharding
> -------------------------------------
>
> Key: BEAM-573
> URL: https://issues.apache.org/jira/browse/BEAM-573
> Project: Beam
> Issue Type: Improvement
> Components: sdk-java-extensions
> Affects Versions: 0.1.0-incubating, 0.2.0-incubating
> Reporter: Jean-Baptiste Onofré
> Assignee: Jean-Baptiste Onofré
> Fix For: Not applicable
>
>
> Currently, TextIO supports sharding (or {{withoutSharding()}} option) only in {{Write}}.
> I think it makes sense to also supports sharding on the {{Read}} behaving like:
> read the input in chunks of n MB (to avoid heap OutOfMemory with large files).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)