Any tips on improving perfomance of the DefaultRunLauncher or using a different one with better perfomance ? At the moment its the bottleneck on what should be a 10 second job. I'm testing on my staging enviroment which is an EC2 instance (2 vpcus , 8gib) results on left. Running on docker. Picture on the right is my macbook pro 16gb M1 , obviously a much more powerful machine but just would like to check if the difference is just machines or docker

Would it still be so different on a dummy operator which would just make

time.sleep(1)

? Sounds like it could be more affected by a network connection than a CPU-bound computations, don't you try to access a data stream (which is near your work network) from a distant AWS region?