Redshift architecture review and storage deep dive

In this section, we will take a deeper dive into the architecture of Redshift clusters, as well as into how data in tables is stored across Redshift nodes. This in-depth look will help you understand and fine-tune Redshift’s performance, though we will also cover how many of the design decisions affecting table layout can be automated by Redshift.

In Chapter 2, Data Management Architectures for Analytics, we briefly discussed how the Redshift architecture uses leader and compute nodes. Each compute node contains a certain amount of compute power (CPUs and memory), as well as a certain amount of local storage. When configuring your Redshift cluster, you can add multiple compute nodes, depending on your compute and storage requirements. Note that to provide fault tolerance and improved durability, the compute nodes have 2.5–3x the stated node storage capacity (for example, if the addressable storage capacity is listed...