Materialize once, query anywhere: Introducing Iceberg materialized views in Amazon Redshift

Amazon Redshift now supports creating and refreshing Iceberg materialized views that pre-compute expensive joins and aggregations once, storing results as standard Apache Iceberg tables in Amazon S3 queryable by any Iceberg-compatible engine.

• Materialized views store pre-computed results as open Iceberg tables in S3, readable by Athena, Spark, SageMaker, and other engines without data copying
• Incremental refresh tracks Iceberg snapshots and recomputes only changed partitions for SUM, COUNT, and non-aggregated queries
• Supported on Amazon Redshift Serverless and RG (Graviton-powered) instances; RA3 and DC2 not supported
• Cross-engine access enables single source of truth; multiple teams query same Iceberg table from S3 at standard read cost
• Use cases include medallion architectures, AI feature stores, cost optimization through compute consolidation, and governed data sharing
• Complements RMS materialized views; use Iceberg MVs for cross-engine access, RMS MVs for lowest read latency from Redshift only

Iceberg materialized views enable organizations to compute analytics once on Redshift and share optimized results across all analytics engines, reducing redundant compute and ensuring metric consistency.