Amazon Redshift is AWS's fully managed cloud data warehouse — a massively parallel processing (MPP) columnar database designed for analytical workloads across terabytes to petabytes of data. Unlike transactional databases optimised for row-level operations, Redshift is optimised for high-throughput aggregation queries that read large fractions of many columns across billions of rows. A query that takes minutes in PostgreSQL on a 500GB table often completes in seconds on Redshift because MPP distributes the work across many nodes and columnar storage reads only the columns needed.
Redshift's architecture separates the leader node from compute nodes. The leader node parses SQL queries, builds execution plans, and coordinates query distribution. Compute nodes execute the plan in parallel, each processing a slice of the data. When a query scans the deliveries fact table, the leader distributes the scan to all compute nodes simultaneously — each node scans its local slice and returns partial results, which the leader aggregates into the final result. The number of compute nodes and instance type determine the cluster's compute capacity and total storage.