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

Posted to issues@aurora.apache.org by "Maxim Khutornenko (JIRA)" <ji...@apache.org> on 2015/04/14 22:06:59 UTC

[jira] [Commented] (AURORA-1219) Improve preemptor efficiency

    [ https://issues.apache.org/jira/browse/AURORA-1219?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14494772#comment-14494772 ] 

Maxim Khutornenko commented on AURORA-1219:
-------------------------------------------

https://reviews.apache.org/r/32907/

> Improve preemptor efficiency
> ----------------------------
>
>                 Key: AURORA-1219
>                 URL: https://issues.apache.org/jira/browse/AURORA-1219
>             Project: Aurora
>          Issue Type: Task
>          Components: Scheduler
>            Reporter: Maxim Khutornenko
>            Assignee: Maxim Khutornenko
>
> While AURORA-1158 makes the preemptor asynchronous and helps scheduling loop performance, it loses some preemption efficiency due to adapting the original algorithm to work asynchronously:
> - Preemptor repeatedly searches for preemption slots with the same ClusterState increasing the probability of victim collisions. 
> - Slave/offer/resource mappings are redundantly recomputed for every pending task slot search.
> - Pending tasks are repeatedly sized up against every slave (as opposed to the other way around) making internal loop optimizations more expensive (i.e. pending task count is expected to be << slave count under normal conditions).



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