Amazon Redshift introduces RG instances with AWS Graviton processors

Amazon Redshift has officially launched RG instances, a new generation of cluster nodes powered by AWS Graviton processors. These new instances promise enhanced performance, with the ability to handle data warehouse and data lake tasks up to 2.4 times faster than the previous RA3 instances, while offering a 30% reduction in cost per vCPU.

The RG instances are equipped with Redshift’s custom-designed vectorized data lake query engine, which effectively processes Apache Iceberg and Parquet data directly on the cluster nodes. This integration allows users to run SQL analytics across both data warehouses and data lakes using a single engine, eliminating the need for the separate Redshift Spectrum scanning fleet and its associated per-terabyte fees.

Whether you’re managing structured data warehouse tasks on Redshift Managed Storage or querying open-format data lake tables on Amazon S3, RG instances offer significant performance boosts. They are up to 2.2 times faster for data warehouse operations, up to 2.4 times faster for Apache Iceberg queries, and up to 1.5 times faster for Parquet workloads.

The newly developed data lake engine includes a specially designed I/O subsystem with features such as smart prefetch, NVMe caching, vectorized Parquet scans, and advanced file and partition-level pruning. The Just-in-Time (JIT) Analyze feature ensures consistently rapid queries by automatically collecting and updating table statistics as data and workload patterns change. Additionally, intelligent NVMe caching keeps frequently accessed datasets near the compute resources, minimizing the need for repeated trips to the data lake and thus speeding up response times.

RG instances are initially available in two sizes: rg.xlarge and rg.4xlarge. Existing RA3 clusters can transition to the new instances using methods such as Snapshot & Restore, Elastic Resize, or Classic Resize. Pricing options include On-Demand and Reserved Instances for one or three years, with no upfront payment required. For full pricing information, users can visit the Amazon Redshift pricing page.

These RG instances are now accessible in various AWS Regions, including US East (N. Virginia and Ohio), US West (Oregon and N. California), Canada (Central), South America (São Paulo), multiple locations in Europe, and a broad range of regions across the Asia Pacific.

For more information on getting started, users can refer to the Amazon Redshift RG Instance Documentation, the RA3 to RG Upgrade Guide, and the Amazon Redshift Pricing page.