Hard to give any specifics without knowing how your data is being
streamed in. If both sources are in the same datastore, you can always
join at the data source and then emit the joined (left or otherwise)
records to your stream processor. On the other hand, if only Stream A or
B is coming in,
So, I have a question related to windowing:
None of the records from A or B are timestamped.
The collections are backed by two master tables (each containing say a few
million records) in a database and I want to use flink to basically join
them and create an enriched stream C.
Hence I won't be a
Hi Sachin,
You should be able to use `coGroup`:
```
streamA.coGroup(streamB)
.where(StreamAFieldFSelector).isEqualTo(StreamBFieldFSelector)
.window(/*your windowing spec*/)
.apply(/*your cogroup function*/)
```
Your coGroup function will look something like:
```
@Override
public void
Hi,
I have two streams A and B.
Which can be joined or connected using a field f.
However, for a given record in A for f = f1, there does not exist any
record in B matching this field f = f1.
In such cases I want to do a left outer join where the combined record
pushed downstream would only have
