Kafka use consumers to deal with the produced data in Queue. Kafka happens at virtual level (like process level) .But the processing speed is somehow restricted where it comes to CPU which is basically a Stack .
So , its Queue vs Stack .

My thought is to transform processor with Queue to that of Stack . And handle all those pointer / virtual function table / paging / segmentation functionalities which is still untouched at Open source level.

Not sure , if its politically correct or technically correct.