[
https://issues.apache.org/jira/browse/FLINK-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14646634#comment-14646634
]
ASF GitHub Bot commented on FLINK-2324:
---------------------------------------
Github user senorcarbone commented on the pull request:
https://github.com/apache/flink/pull/937#issuecomment-126067262
That would be good ^^
then it's :+1: from me, at least for now.
It's generally good performance-wise to have less serialised states. This
means that we will have a constant number of issued writes to external storage
(== #subtasks). On the other hand this also makes our life harder a bit when it
comes to repartitioning, as you already mentioned we need to revisit this.
> Rework partitioned state storage
> --------------------------------
>
> Key: FLINK-2324
> URL: https://issues.apache.org/jira/browse/FLINK-2324
> Project: Flink
> Issue Type: Improvement
> Reporter: Gyula Fora
> Assignee: Gyula Fora
>
> Partitioned states are currently stored per-key in statehandles. This is
> alright for in-memory storage but is very inefficient for HDFS.
> The logic behind the current mechanism is that this approach provides a way
> to repartition a state without fetching the data from the external storage
> and only manipulating handles.
> We should come up with a solution that can achieve both.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
