Amazon Redshift is a petabyte-scale columnar MPP (Massively Parallel Processing) data warehouse. Today we cover data distribution, sort keys, RA3/Spectrum architecture, and workload optimization through WLM and concurrency scaling.
A Redshift cluster consists of one leader node and multiple compute nodes. The leader node parses queries, plans them, and distributes work to compute nodes. Each compute node further divides into multiple slices, which are the minimum parallelization unit. Data is distributed across slices, and each slice processes its data in parallel.