Hi Team,


Could you please assist with below questions.

1)If apache beam is integrated with flink and then portable runner(flink or 
spark ae used)used what are the overheads?

2)Robert could you please explain in detail regarding below question once (This 
question was asked in my previous conversation)

> if we integrate apache beam with flink and then use apache flink runner what 
> are the overheads?

This is generally pipeline-specific, but there can be some overhead
from using the windowing semantics in Beam. When running portably,
there is also overhead in sending elements over GRPC to the SDK
workers which are run in separate processes (generally docker
containers) than the Flink worker (though this is eliminated using the
legacy runner or embedded mode for portability).


Thanks
Azhar

Reply via email to