Amazon Redshift is AWS's cloud data warehouse. It is a columnar, MPP (massively parallel processing) SQL engine for analytics. You load curated data (often from S3) and run BI / SQL workloads.
S3 / Glue / DMS --> Redshift --> BI tools / dbt / notebooks
(leader + compute nodes)How Redshift feels
- Leader node plans queries; compute nodes store slices of data
- Classic clusters are provisioned (node types / count); Redshift Serverless removes capacity planning
- Spectrum / external tables can query S3 lake data
- Distribution styles (KEY, EVEN, ALL) and sort keys matter for join performance
Redshift vs Snowflake (fresher comparison)
| Topic | Redshift | Snowflake | |---|---|---| | Home cloud | AWS-native | Multi-cloud (AWS/GCP/Azure) | | Architecture | Cluster (or Serverless) | Separated storage + warehouses (compute) | | Scaling | Resize cluster / concurrency scaling | Instant warehouse size / multi-cluster | | Storage | Managed in cluster or RA3 managed storage | Cloud object storage under the hood | | Pricing feel | Node hours or RPU (serverless) | Credits for compute + storage | | Ecosystem | Deep AWS (IAM, S3, Glue) | Strong cross-cloud + Time Travel / Zero-Copy Clone |
When teams pick Redshift
Already all-in on AWS, want Spectrum + Glue Catalog proximity, or existing Redshift skill/cost commitments.
When teams pick Snowflake
Multi-cloud, strong separation of storage/compute, easy clone/time-travel workflows, or vendor neutrality.
Interview tip: Define Redshift as AWS MPP warehouse, mention leader/compute and S3 load path, then contrast Snowflake's storage/compute split and multi-cloud story without dumping feature lists.