[
https://issues.apache.org/jira/browse/FLINK-2237?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15141192#comment-15141192
]
ASF GitHub Bot commented on FLINK-2237:
---------------------------------------
Github user fhueske commented on the pull request:
https://github.com/apache/flink/pull/1517#issuecomment-182483003
The benefit of having multiple partitions is that you do not need to go
over the complete RecordArea but only over a partition. After the partition is
compacted or was emitted, it's MemorySegments can be used by other Partitions
as well. Depending on the strategy to choose the Partition to emit or compact
(#segments in partition, update count on partition, #unique values in
partition), this should improve the combine rate (less emitted records from the
table) and also improve the pipelined behavior of the combiner, because you'll
get more but much shorter pauses.
> Add hash-based Aggregation
> --------------------------
>
> Key: FLINK-2237
> URL: https://issues.apache.org/jira/browse/FLINK-2237
> Project: Flink
> Issue Type: New Feature
> Reporter: Rafiullah Momand
> Assignee: Gabor Gevay
> Priority: Minor
>
> Aggregation functions at the moment are implemented in a sort-based way.
> How can we implement hash based Aggregation for Flink?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
