Amazon Redshift has long delivered cloud data warehouse power at a fraction of on-premises cost. Now, the introduction of RG instances powered by AWS Graviton marks a new generation of performance and efficiency. These instances combine a faster, more cost-effective compute engine with a built-in data lake query engine, enabling you to run SQL analytics across both warehouse tables and Amazon S3 data lakes from a single system. Designed for modern workloads—from BI dashboards to agentic AI—RG instances deliver up to 2.2x faster performance and 30% lower price per vCPU compared to RA3 instances. Below, we answer key questions about this new instance family and how it can transform your analytics.
1. What are Amazon Redshift RG instances and how do they differ from RA3 instances?
Amazon Redshift RG instances are the latest addition to the Redshift instance family, built on AWS Graviton processors—the same custom Arm-based chips used across many AWS services for superior price-performance. They succeed and improve upon the previous RA3 instance family. While RA3 instances already offered managed storage with Redshift Managed Storage (RMS), RG instances take it further: they deliver up to 2.2x faster data warehouse workload performance at a 30% lower price per vCPU. The key differentiator is the integrated data lake query engine, which is enabled by default. This engine lets you query both structured warehouse tables and data lakes (Apache Iceberg and Apache Parquet) using a single SQL engine, eliminating the need for separate query systems. RG instances are also available in sizes like rg.xlarge and rg.4xlarge, which replace legacy RA3 sizes with improved vCPU and memory ratios.
2. What performance improvements do RG instances offer for data warehouse workloads?
RG instances deliver a notable performance leap across multiple dimensions. For standard data warehouse workloads—such as BI dashboards, ETL pipelines, and near-real-time analytics—they run queries up to 2.2x faster than RA3 instances. This is achieved through the combination of Graviton processors and architectural optimizations. Additionally, for low-latency SQL queries, Redshift previously improved new query performance by up to 7x (announced March 2026), further boosting response times. When querying data lakes, the integrated engine shows even more dramatic gains: performance up to 2.4x faster for Apache Iceberg and up to 1.5x faster for Apache Parquet compared to RA3. These improvements are critical for workloads that demand high query volumes and low latency, including autonomous AI agents that can dwarf typical human query usage.
3. How does the integrated data lake query engine benefit analytics?
The integrated data lake query engine is a game-changer for organizations that store data in both Amazon Redshift warehouse tables and Amazon S3 data lakes. Traditionally, querying across these two environments required separate engines, leading to complexity, data movement, and higher costs. With RG instances, the engine is built into the warehouse—no additional setup needed. You can run SQL analytics across both your data warehouse and data lake from a single query engine. For example, you can join a frequently accessed customer table in Redshift with raw IoT data stored as Apache Iceberg files in S3. This not only simplifies operations (one system to manage) but also reduces total analytics costs because you avoid duplicating data or using multiple query tools. The engine is optimized for Parquet and Iceberg formats, delivering up to 2.4x speed improvements over RA3 for Iceberg queries.
4. How do RG instances reduce costs compared to RA3 instances?
RG instances cut costs primarily through a 30% lower price per vCPU compared to RA3 instances, while also delivering higher performance. This means you get more work done per dollar. For example, if you migrate from a ra3.4xlarge (12 vCPU, 96 GB) to an rg.4xlarge (16 vCPU, 128 GB), you get 1.33x the compute and memory for a lower per-vCPU price. The improved performance (up to 2.2x faster) further reduces the time your clusters run, potentially lowering overall costs. Additionally, because the integrated data lake query engine eliminates the need for separate query services or data movement, you save on associated compute and storage costs. For mixed workloads (data warehouse + data lake), using RG instances can lower your total analytics spend. To estimate your savings, AWS recommends using the AWS Pricing Calculator with your specific workload patterns.
5. What use cases are best suited for RG instances?
RG instances are designed for any workload demanding high performance, low latency, and cost efficiency—especially those driving high query volumes. Key use cases include:
- Business Intelligence (BI) dashboards – Faster refresh times for interactive reports.
- ETL pipelines – Accelerated data transformation and loading.
- Near-real-time analytics – Low-latency SQL for operational insights.
- Agentic AI workloads – Autonomous AI agents that query the warehouse at massive scale; RG instances handle the load without spiraling costs.
- Unified warehouse/data lake analytics – Combining structured tables with S3 data lakes using a single engine.
For small departmental analytics, the rg.xlarge (4 vCPU, 32 GB) is ideal. For standard production with medium data volumes, the rg.4xlarge (16 vCPU, 128 GB) is recommended. The blend of speed, cost, and integrated lake query makes RG instances a future-proof choice.
6. How can customers migrate or start using RG instances?
Getting started with Amazon Redshift RG instances is straightforward. You can launch new clusters or migrate existing clusters through the AWS Management Console, AWS Command Line Interface (AWS CLI), or AWS API. The integrated data lake query engine is enabled by default, so no additional configuration is needed. AWS provides a comparison table to help you map current RA3 instances to recommended RG sizes (e.g., ra3.xlplus → rg.xlarge, ra3.4xlarge → rg.4xlarge). For existing clusters, you can modify the instance type via the console; automation scripts can handle bulk migrations. AWS recommends using the AWS Pricing Calculator to estimate savings based on your specific workload patterns before migrating. There is no additional cost to enable the data lake query engine—it’s included with the instance.
7. How does the instance comparison table map RA3 to RG?
The official migration guide provides a direct mapping from current RA3 instances to the recommended RG instance, ensuring you get optimal performance and cost benefits. Here’s the key comparison:
- ra3.xlplus (4 vCPU, 32 GB) → rg.xlarge (4 vCPU, 32 GB) – identical specs, ideal for small cluster departmental analytics.
- ra3.4xlarge (12 vCPU, 96 GB) → rg.4xlarge (16 vCPU, 128 GB) – a 1.33:1 increase in both vCPU and memory, suited for standard production workloads with medium data volumes.
This scaling provides immediate performance gains and cost savings. Note that for larger RA3 sizes not listed above, AWS likely offers corresponding RG equivalents; check the documentation. The AWS Pricing Calculator can also help you choose the right RG size for your specific needs.