[
https://issues.apache.org/jira/browse/FLINK-38810?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Roman Khachatryan updated FLINK-38810:
--------------------------------------
Description:
On timer, TimeIntervalJoin potentially performs a lot of work iterating through
all it's records in state.
And when multiple timers are triggered by a single watermark, checkpoint can be
delayed significantly, effectively increasing e2e latency.
Recently, splittable timers were introduced to Flink for a limited set of
operators in FLINK-20217.
TimeIntervalJoin can benefit from it as well.
was:
On timer, TimeIntervalJoin potentially performs a lot of work iterating through
all it's records in state.
And when multiple timers are triggered by a single watermark, checkpoint can be
delayed significantly, effectively increasing e2e latency.
Recently, splittable timer functionality was introduced to Flink for a limited
set of operators.
TimeIntervalJoin can benefit from it as well.
> Use splittable timers in Interval Join
> --------------------------------------
>
> Key: FLINK-38810
> URL: https://issues.apache.org/jira/browse/FLINK-38810
> Project: Flink
> Issue Type: Improvement
> Components: Table SQL / Runtime
> Affects Versions: 2.2.0
> Reporter: Roman Khachatryan
> Assignee: Roman Khachatryan
> Priority: Major
> Fix For: 2.2.1
>
>
> On timer, TimeIntervalJoin potentially performs a lot of work iterating
> through all it's records in state.
> And when multiple timers are triggered by a single watermark, checkpoint can
> be delayed significantly, effectively increasing e2e latency.
> Recently, splittable timers were introduced to Flink for a limited set of
> operators in FLINK-20217.
> TimeIntervalJoin can benefit from it as well.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
