Primitive

Data Warehouse

A columnar analytical store optimised for large-scale aggregations and BI queries over historical data.

Description

The cloud warehouses (AWS Redshift, GCP BigQuery, Azure Synapse) are all columnar, massively-parallel analytical stores, but their architectures differ notably. BigQuery is fully serverless and bills by data scanned, Redshift has traditionally been cluster/node-provisioned (with a serverless option added later), and Synapse combines provisioned and serverless SQL pools with Spark-based analytics. They also differ in how storage and compute separate, in streaming-ingestion support, and in their BI and data-lake integrations. To the best of our knowledge all serve large-scale analytics well; the choice hinges on the provisioning/billing model and how it fits your query patterns.

Capabilities

  • Columnar storage for fast aggregation
  • Massively parallel query execution
  • Separate storage and compute scaling
  • Batch and streaming ingestion

Vendor implementations

Icon-Architecture/64/Arch_Amazon-Redshift_64
AWS Redshift
AZ Synapse
Icon_24px_BigQuery_Color
GCP BigQuery

Related primitives