Where the rest of the estate already runs on AWS, Redshift removes a lot of movement: it reads directly from S3, shares data across clusters without copying, and sits inside the same IAM model as everything else. The trade is that distribution and sort keys are decisions you make early and feel for years.
Overview
Distribution keys are not a detail
Redshift performance is decided mostly by how data is distributed across nodes and how it is sorted within them. Get it right and joins happen locally; get it wrong and every query redistributes data across the network. We model this against the joins you actually run rather than the ones the schema implies.
What we deliver
Serverless or provisioned
Redshift Serverless removes cluster management and suits variable workloads; provisioned still wins on sustained heavy use and reserved pricing. The right answer depends on your load profile, and it is worth measuring rather than assuming.
- Distribution and sort key design against real join patterns
- Spectrum for querying S3 without loading it
- Data sharing across clusters instead of copies
- Workload management queues so reporting survives a heavy load
- Serverless or provisioned, chosen on measured load
Why it matters
Why teams pick it
Redshift is the pragmatic choice when the data already lives in S3 and the organisation is committed to AWS, because it removes egress, duplication and a second identity model. It is less compelling as a standalone warehouse bought on its merits alone.
Workflow
How we work in it.
- 01
Assessment
- Existing AWS estate and where the data sits
- Join and query patterns
- Load profile: steady, bursty or seasonal
- Compliance and access boundaries
- 02
Cluster & Schema Design
- Serverless or provisioned, with the reasoning
- Distribution and sort key strategy
- Spectrum external tables where loading is unnecessary
- Workload management queues
- 03
Build & Load
- Ingestion from S3, streams and operational databases
- Layered modelling with tested transformations
- Data sharing rather than duplication
- Vacuum, analyse and maintenance scheduling
- 04
Operate
- Query monitoring and queue tuning
- Cost review and reserved capacity planning
- Access review and audit
- Handover and training
Related in Warehouse & lakehouse
