Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
For windowed KStream / KTable in Streams DSL, the library will use a WindowedStreamPartitioner dynamically at runtime. But this configuration is relying on runtime type inference which is not efficient. Another issue is that for dynamic routing on sink node, the topic name would not be known before hand so the partitioner need to call a different partition function that requires a topic name as parameter.
We can consider if there is a better way to configure the windowed stream partitioner when making windowed KStream / KTable as its own secondary class in the DSL.