In a scenario where you need to perform complex data aggregations on a large dataset stored in Amazon Redshift, what strategies would you employ to optimize the performance of these aggregations? Consider the use of indexes, distribution styles, and sorting keys.