[
https://issues.apache.org/jira/browse/NIFI-15091?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kevin Doran updated NIFI-15091:
-------------------------------
Description:
There was recently an [interesting
discussion|https://github.com/apache/nifi/pull/10372#issuecomment-3362326306]
on the use of {{-XX:+UseStringDeduplication}} benefiting NiFi clusters that
process many FlowFiles containing many FlowFile attributes, potentially with
duplicate values.
At a glance, it seems that many NiFi use cases would benefit from enabling the
{{UseStringDeduplication}} JVM GC arg, and there is probably little to no
downside or risk to enabling it by default, even for use cases might not
benefit.
We should profile one or more representative flows in NiFi with and without
this flag, quantify the potential savings, and, if merited, consider enabling
this GC option by default in the JVM args of {{bootstrap.conf}} sourced by
{{nifi.sh}}.
was:
There was recently an [interesting
discussion|https://github.com/apache/nifi/pull/10372#issuecomment-3362326306]
on the use of `-XX:+UseStringDeduplication` benefiting NiFi clusters that
process many FlowFiles containing many FlowFile attributes, potentially with
duplicate values.
At a glance, it seems that many NiFi use cases would benefit from enabling the
`UseStringDeduplication` JVM GC arg, and there is probably little to no
downside or risk to enabling it by default, even for use cases might not
benefit.
We should profile one or more representative flows in NiFi with and without
this flag, quantify the potential savings, and, if merited, consider enabling
this GC option by default in the JVM args of `bootstrap.conf` sourced by
`nifi.sh`.
> Evaluate NiFi's default JVM GC args. Consider adding UseStringDeduplication
> ---------------------------------------------------------------------------
>
> Key: NIFI-15091
> URL: https://issues.apache.org/jira/browse/NIFI-15091
> Project: Apache NiFi
> Issue Type: Improvement
> Components: Configuration
> Reporter: Kevin Doran
> Assignee: Kevin Doran
> Priority: Minor
>
> There was recently an [interesting
> discussion|https://github.com/apache/nifi/pull/10372#issuecomment-3362326306]
> on the use of {{-XX:+UseStringDeduplication}} benefiting NiFi clusters that
> process many FlowFiles containing many FlowFile attributes, potentially with
> duplicate values.
> At a glance, it seems that many NiFi use cases would benefit from enabling
> the {{UseStringDeduplication}} JVM GC arg, and there is probably little to no
> downside or risk to enabling it by default, even for use cases might not
> benefit.
> We should profile one or more representative flows in NiFi with and without
> this flag, quantify the potential savings, and, if merited, consider enabling
> this GC option by default in the JVM args of {{bootstrap.conf}} sourced by
> {{nifi.sh}}.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
