Meta interview question

How would you handle processing large datasets in a constrained environment?