You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@beam.apache.org by "Robert Burke (JIRA)" <ji...@apache.org> on 2019/01/07 17:30:00 UTC

[jira] [Commented] (BEAM-6373) Make Go SDK on Dataflow multi-threaded

    [ https://issues.apache.org/jira/browse/BEAM-6373?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16736082#comment-16736082 ] 

Robert Burke commented on BEAM-6373:
------------------------------------

Without SplitableDoFn (SDF) , Go SDK jobs are unable to scale and divide work, which includes any kind of Parallelism. This is being worked on this quarter. See [https://beam.apache.org/roadmap/go-sdk] for additional information.

> Make Go SDK on Dataflow multi-threaded
> --------------------------------------
>
>                 Key: BEAM-6373
>                 URL: https://issues.apache.org/jira/browse/BEAM-6373
>             Project: Beam
>          Issue Type: Improvement
>          Components: runner-dataflow, sdk-go
>            Reporter: Andrew Brampton
>            Assignee: Robert Burke
>            Priority: Major
>
> When running on dataflow, with a machine with more than 1 V-CPU, the worker will never use more than 1 core of CPU. I presume this is because the Dataflow runner does not allow the worker to run multiple steps concurrently.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)