Hi,
in general, I think it is clearly a good idea to make rescaling as cheap and as
dynamic as possible and this is a goal that is surely on the long term roadmap
of Flink.
However, from a technical point of view I think that things are not as simple
if you go into details, and details is what
Currently, the rescale operation is to stop the whole job and restart it
with different parrellism. But the rescale operation cost a lot and took
lots of time to recover if the state size is quite big.
And a long-time rescale might cause other problems like latency increase
and back pressure. For
