Amazon Redshift
Analytics data warehouse
A columnar, massively parallel (MPP) data warehouse for running analytical SQL queries on large datasets; includes Redshift Spectrum to query data directly on S3.
- Cost$$$ High
- Difficulty●●●●●
- Costs money when idle?Yes
- Managed by AWS
- In examsCLF · SAA · SOA
Think of it asLike a warehouse organized to quickly answer summary questions, not to sell items one at a time.
When to use it
✓ Use it when
BI, reporting, and OLAP analysis on terabytes to petabytes of data.
✗ Avoid it when
Day-to-day OLTP transactions → RDS / Aurora; occasional queries → Athena.
Common uses
- BI dashboards for leadership
- Analyze years of sales data
- Combine with an S3 data lake via Spectrum
What does it cost?
- Pricing model: Provisioned: per node-hour (RA3 adds managed storage); Spectrum by data scanned.
Provisioned clusters are billed even with no queries — you can pause them or use Reserved Nodes.
When the question says data warehouse, OLAP on TB to PB, BI and complex queries, choose Redshift; to query S3 data directly from Redshift, Redshift Spectrum; for ad-hoc with nothing to set up, Athena; for OLTP, RDS or Aurora.
Frequently asked questions
How is Redshift different from RDS or Aurora?
RDS and Aurora are transactional (OLTP) databases: many small row-level reads and writes. Redshift stores data by column and processes in parallel (MPP), optimized for analytical (OLAP) queries that scan and aggregate large volumes. Using Redshift for a transactional application is inefficient, and the reverse is also true.
What are Redshift Spectrum and Redshift Serverless?
Spectrum lets Redshift query data directly in S3 without loading it into the cluster, suited to cold or very large data. Redshift Serverless removes choosing and managing a cluster: AWS provisions capacity and charges by usage. Both help separate compute from storage.
How do loading, backup and fault tolerance work in Redshift?
The efficient way to load is the COPY command from S3, which runs in parallel, rather than row-by-row INSERT. Redshift takes snapshots automatically and can copy them to another Region for DR. Multi-AZ configuration protects against an AZ failure, and data can be encrypted with KMS.
Related services
SAA-C03 study guide
Redshift is covered in these study modules: