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

Posted to issues@drill.apache.org by "Steven Phillips (JIRA)" <ji...@apache.org> on 2014/11/24 19:21:12 UTC

[jira] [Updated] (DRILL-1370) Improve fragment staging

     [ https://issues.apache.org/jira/browse/DRILL-1370?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Steven Phillips updated DRILL-1370:
-----------------------------------
    Fix Version/s:     (was: 0.7.0)
                   Future

> Improve fragment staging
> ------------------------
>
>                 Key: DRILL-1370
>                 URL: https://issues.apache.org/jira/browse/DRILL-1370
>             Project: Apache Drill
>          Issue Type: Improvement
>          Components: Execution - Flow
>            Reporter: Jacques Nadeau
>            Assignee: Steven Phillips
>             Fix For: Future
>
>         Attachments: DRILL-1370.patch
>
>
> Drill optimistically executes all major fragments of a query. In some situations, this greedy behavior is too optimistic, which causes portions of the query to swamp the IO path too early, starving earlier phases of the query.  Our RPC backpressure should help manage this.  However, with a high level of parallelization, the back pressure is too lenient. 



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