If you mean time-windowed join documented here [1]. I think it implicitly uses keyed stream [2] where the key is the field in equi-join predicate. The window state is also keyed [3] in this case. I also cc Timo and Piotr, they might add more to this topic.
[1] https://ci.apache.org/projects/flink/flink-docs-master/dev/table/sql.html#joins [2] https://github.com/apache/flink/blob/8674b69964eae50cad024f2c5caf92a71bf21a09/flink-libraries/flink-table/src/main/scala/org/apache/flink/table/plan/nodes/datastream/DataStreamWindowJoin.scala#L261 [3] https://github.com/apache/flink/blob/c8b2ee27d80b1437dd65a9327da65c251febd736/flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/join/TimeBoundedStreamJoin.scala#L78 On Mon, Dec 24, 2018 at 6:29 PM Anil <anilsingh....@gmail.com> wrote: > Thanks for the quick response Andrey. I'm doing a SQL time-windowed join on > non-keyed stream. > So all the thread in various task slot in the same TM will share this > state. > > > > -- > Sent from: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/ >