Amazon Redshift

Amazon Redshift is AWS's managed data warehouse service, designed for running complex analytical queries across large datasets using a massively parallel processing (MPP) architecture. Originally launched in 2013, it was one of the first cloud data warehouses to make petabyte-scale analytics economically accessible.

Provisioned clusters let you choose node types (RA3 recommended) and scale by adding nodes. RA3 nodes decouple storage from compute using Amazon S3-backed managed storage, so you pay for compute capacity separately from storage and can scale each independently. Legacy DC2 nodes use local SSD storage for smaller, compute-intensive workloads.

Redshift Serverless (launched 2022) removes cluster management entirely. You pay per RPU (Redshift Processing Unit) per second while queries are running. At $0.375/RPU-hour, it auto-scales capacity up and down based on workload demand, making it cost-efficient for intermittent analytics.

Redshift Spectrum extends queries to data stored in Amazon S3 in open formats (Parquet, ORC, JSON) without loading it into Redshift. You pay $5/TB scanned, enabling a data lake + data warehouse architecture on top of S3.

AQUA (Advanced Query Accelerator) is hardware-accelerated caching that pushes query processing closer to storage, delivering up to 10x faster performance for compute-intensive queries on RA3 nodes at no additional cost.

Federated queries allow Redshift to query data in RDS PostgreSQL, RDS MySQL, and Aurora in real time, combining transactional data with analytical data in a single query.

Redshift ML integrates with Amazon SageMaker to create and run ML models from SQL, similar to BigQuery ML.

Redshift uses PostgreSQL-based SQL dialect — most PostgreSQL syntax works, though some features differ. This familiarity reduces the learning curve for teams already using PostgreSQL.

Deep AWS ecosystem integration with S3, Glue, Lake Formation, QuickSight, SageMaker, and IAM makes Redshift the natural choice for analytics-heavy AWS workloads.