I am reading ray 1.0 architecture document and this is part 2 of my questions.
Question 1:
1.a - In the screenshot below, what/who is the owner?
For example, if calculation x relies on calculation y, and calculation y relies on calculation z. Then x owns y and z, and y owns z ? So x is the main mainer, while y is considered a sub-owner? Does this mean the term owner is dynamic based on what workers depend on it?
1.b - Can the driver also be the owner? Or only workers can be owners?
Question 2:
2.a - What is the difference between the worker in the head node and the worker at the driver node?
2.b - Why is there no connection between the driver and the worker at the head node in the figure below? Do they not communicate or does that worker in the head node serve a different purpose?
Question 3:
3.a - Does the head node always have 1 driver and 1 worker? Or can it have 1 driver and n amount of workers?
3.b - If n amount of workers what dictates that number?
Question 4:
4.a - Does the worker at the head node have an ownership table and an in-process store? like the other workers in the worker node?