[
https://issues.apache.org/jira/browse/FLINK-10432?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16739120#comment-16739120
]
shuai.xu commented on FLINK-10432:
----------------------------------
Hi [~srichter], it's really a good idea. I think we could benefit more from it.
With this work we could support Concurrent Group based scheduling strategy
easily to satisfy the flexible requirements for different jobs, such as the
FLINK-10240. The idea of Concurrent Group based scheduling is in FLINK-11298.
> Introduce bulk/group-aware scheduling
> -------------------------------------
>
> Key: FLINK-10432
> URL: https://issues.apache.org/jira/browse/FLINK-10432
> Project: Flink
> Issue Type: Sub-task
> Components: Distributed Coordination
> Affects Versions: 1.7.0
> Reporter: Stefan Richter
> Assignee: Stefan Richter
> Priority: Major
>
> In this step, we change our new Scheduler to support reasoning over a bulk of
> Executions before interacting with the SlotPool instead of requesting slots
> on a per-execution basis. This gives us the opportunity to group and schedule
> tasks together as one unit and to apply scheduling algorithms that require a
> more holistic view on the tasks at hand. For now, this step will probably
> mainly address streaming / EAGER scheduling. After this step, we expect that
> we can support the use-case of local recovery.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
