[
https://issues.apache.org/jira/browse/FLINK-13792?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Flink Jira Bot updated FLINK-13792:
-----------------------------------
Priority: Minor (was: Major)
> source and sink support manual rate limit
> -----------------------------------------
>
> Key: FLINK-13792
> URL: https://issues.apache.org/jira/browse/FLINK-13792
> Project: Flink
> Issue Type: Improvement
> Components: Connectors / Common
> Affects Versions: 1.8.1
> Reporter: zzsmdfj
> Priority: Minor
> Labels: auto-deprioritized-major
>
> in current flink implement automatic flow control by back pressure, it is
> efficient for the most scene, but in some special scenario, do we need
> fine-grained flow control to avoid impact on other systems? For example: if i
> have window with days(a lot of datas), then do call ProcessWindowFunction
> when trigger, this will produce a lot of data to sink, if sink to message
> queue, it can have a huge impact to message queue. so if there is sink rate
> limiter, it is friendly to external system. for source rate limiter, it is
> appropriate for having window operator and accumulating a large amount of
> historical data.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
