[
https://issues.apache.org/jira/browse/FLINK-15455?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Flink Jira Bot updated FLINK-15455:
-----------------------------------
Labels: stale-major (was: )
I am the [Flink Jira Bot|https://github.com/apache/flink-jira-bot/] and I help
the community manage its development. I see this issues has been marked as
Major but is unassigned and neither itself nor its Sub-Tasks have been updated
for 30 days. I have gone ahead and added a "stale-major" to the issue". If this
ticket is a Major, please either assign yourself or give an update. Afterwards,
please remove the label or in 7 days the issue will be deprioritized.
> Enable TCP connection reuse across multiple jobs.
> -------------------------------------------------
>
> Key: FLINK-15455
> URL: https://issues.apache.org/jira/browse/FLINK-15455
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Network
> Reporter: Yingjie Cao
> Priority: Major
> Labels: stale-major
>
> Currently, tcp connections can be only reuse by tasks residing in the same
> TaskManager and consumes the same IntermediateResult. And after job finish or
> failover, the TCP connections are closed and new connections must be setup
> latter.
> As an improvement, we can make tcp connections a cluster level resource which
> can be reused by multi jobs. The advantages are as follows:
> # Reduce the number of TCP connections so we can save some resources.
> # Reduce the overhead of connection setup and close so restarted jobs after
> failover and latter jobs submitted to the same session cluster can reuse the
> previous connections.
> We use Flink session cluster as a service for ad-hoc queries and the users
> can produce some statistics or create some statements and reports at any
> time. Most of the queries finish in 2s and we find tcp connection reuse help
> a lot to reduce the average execution time which means more queries can be
> processed using the same resource and time with even better user experience.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
