Your question is Optimizing Skewed Joins. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
Given two large tables with highly skewed distribution keys, how would you optimize a complex SQL JOIN operation to avoid performance bottlenecks for Roku-scale datasets?