[ 
https://issues.apache.org/jira/browse/FLINK-22748?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17886904#comment-17886904
 ] 

Benoît Paris commented on FLINK-22748:
--------------------------------------

Well, I'd like to pause and congratulate all the contributors!

First [~twalthr] , [~jark] for the issue management and reviews, 
[~nicholasjiang] for the first attempt, then [~klam-shop] for picking up the 
feature again; and finally [~arvid] for the numerous reviews helping this 
getting merged.

Here we are 3 years later; and I can finally enjoy this quite niche feature. 
For free thanks to you guys.

I'm quite curious as to its use other than the one I have, now. I'll share it:

This is for distributing reactive web apps. Backend publishes messages from a 
user to a topic, business rules and event mixing happen in Flink, back it goes 
to the backend for serving the emergent results to potentially other users. 
Before this patch I had to make backends read all the messages of a single 
topic and filter, now they'll register which topic they're going to listen to, 
and which user they are serving; and Flink will make the routing happen, making 
a part of my system scalable.

So, what's yours at Celeborn, [~nicholasjiang] ? And yours at Shopify 
[~klam-shop] ?

 

> Allow dynamic target topic selection in SQL Kafka sinks
> -------------------------------------------------------
>
>                 Key: FLINK-22748
>                 URL: https://issues.apache.org/jira/browse/FLINK-22748
>             Project: Flink
>          Issue Type: Improvement
>          Components: Connectors / Kafka, Table SQL / Ecosystem
>            Reporter: Timo Walther
>            Assignee: Kevin Lam
>            Priority: Major
>              Labels: pull-request-available, stale-assigned, starter
>             Fix For: kafka-3.3.0
>
>
> We should allow to write to different Kafka topics based on some column value 
> in SQL.
> The existing implementation can be easily adapted for that. The "target 
> topic" would be an additional persisted metadata column in SQL terms. All one 
> need to do is to adapt
> DynamicKafkaSerializationSchema
> KafkaDynamicSink
> We should guard this dynamic behavior via a config option and make the topic 
> option optional in this case.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to