How would you optimize a query that is joining a 10-billion row fact table with a heavily skewed dimension table for Arlo analytics?
