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

Posted to dev@tinkerpop.apache.org by "Marko A. Rodriguez (JIRA)" <ji...@apache.org> on 2015/08/27 16:36:45 UTC

[jira] [Commented] (TINKERPOP3-705) Make use of a MemoryRDD instead of Accumulators in SparkGraphComputer?

    [ https://issues.apache.org/jira/browse/TINKERPOP3-705?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14716773#comment-14716773 ] 

Marko A. Rodriguez commented on TINKERPOP3-705:
-----------------------------------------------

When we bump to 1.4.0 Spark, lets see if their Accumulator has evolved. If not, then do a MemoryRDD. This should be relatively simple to do, but hopefully, Spark's accumulator is solid now.

> Make use of a MemoryRDD instead of Accumulators in SparkGraphComputer?
> ----------------------------------------------------------------------
>
>                 Key: TINKERPOP3-705
>                 URL: https://issues.apache.org/jira/browse/TINKERPOP3-705
>             Project: TinkerPop 3
>          Issue Type: Improvement
>          Components: hadoop
>    Affects Versions: 3.0.0-incubating
>            Reporter: Marko A. Rodriguez
>            Assignee: Marko A. Rodriguez
>             Fix For: 3.1.0-incubating
>
>
> The current problem with accumulators in Spark is that they must be in an action, not midway in an RDD computation. A {{MemoryRDD}} can simulate such aggregation. Think it through and see if makes sense for SparkGraphComputer.



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