Amazon Redshift: The Definitive Guide

Chapter 5. Scaling and Performance Optimizations

If we told you that the only constant is change, then most likely we would be “preaching to the choir.” The challenge today is how fast your data warehouse can adapt to the change. With traditional data warehousing systems, this change is often difficult because of lead time to provision resources. With Amazon Redshift, adapting to change is easy, be it changes in storage needs or changes in compute needs. There are no expensive wrong decisions as you can quickly scale with the increase or decrease in demand.

The objective of scaling is to meet changes in your workload to maintain current performance levels and associated SLA. If you add new workloads to your warehouse, then existing workload SLAs can get impacted; this is where scaling comes in. Scaling could also be required if you are analyzing more data than before, which has caused a visible impact to your workload SLAs. To achieve your scaling goals using Amazon Redshift, there are two strategies to consider: ensuring your data warehouse is sized correctly and ensuring that your workloads are tuned for performance.

With Amazon Redshift, you can size your data warehouse by scaling the compute vertically as well as horizontally (see Figure 5-1). Vertical scaling is when you scale “up” by having additional compute that is operating on a single query. Scaling up results in the total number of vCPUs or memory increasing. If you need to retain the SLAs of existing workloads and ...