How does Redshift RA3 node architecture separate compute and storage, and why does this matter?

Redshift RA3 nodes use managed storage that automatically scales independently of compute capacity, meaning you can scale your compute power up or down for performance needs without needing to also move or resize your actual stored data, since RA3 uses high performance SSDs as a local cache in front of data that is actually stored durably in Amazon S3 behind the scenes, giving you far more flexibility than older node types where compute and storage were tightly coupled together.

aws redshift resize-cluster --cluster-identifier my-cluster --node-type ra3.xlplus --number-of-nodes 4