Redshift
Amazon Redshift is AWS’s data warehouse: a columnar, massively parallel SQL engine built for analytical queries over large volumes of historical data. It is the OLAP half of the pairing described in OLTP vs OLAP — the place reporting queries go so that they do not run against the transactional database.
Two deployment models exist, and new work almost always starts with the first:
- Redshift Serverless, generally available since July 2022, has no cluster to size or manage. You set a base capacity in Redshift Processing Units (RPUs, one RPU providing 16 GB of memory), the warehouse scales itself with the workload, and it costs nothing between queries.
- Provisioned clusters give explicit control of node type and count, and remain the right choice for steady, predictable warehouse workloads where reserved capacity can be bought against them.
The two share almost all their features: snapshots, materialized views, stored
procedures, federated queries, data sharing, semi-structured SUPER data, spatial
functions, machine learning and querying across databases.
The pages here
Section titled “The pages here”- Amazon Redshift overview — architecture, columnar storage, Redshift Spectrum and data lake integration.
- Redshift in an analytics architecture — where the warehouse sits between ingestion, processing and consumption.
- Redshift’s ACID compliance — which guarantees it actually offers, and where its isolation model differs from an OLTP database.