Amazon Redshift RG Instances: Graviton-Powered Performance and Unified Data Lake Querying

Introduction

Since its launch in 2013, Amazon Redshift has revolutionized cloud data warehousing by delivering enterprise-grade performance at a fraction of on-premises costs. Over the years, each new generation—from dense compute to RA3 instances, from provisioned to serverless—has driven down query costs while boosting speed and efficiency. Today, as data volumes surge and analytics demands become more complex, organizations need solutions that can handle both structured warehouse data and diverse data lake datasets without breaking the bank. Enter the new Amazon Redshift RG instances, powered by AWS Graviton processors, which promise a leap forward in performance, cost savings, and integrated data lake querying.

The Evolution of Amazon Redshift

For more than a decade, Amazon Redshift has continuously evolved to meet changing data analytics needs. Businesses now frequently combine structured tables in the warehouse with cost-effective data lakes on Amazon S3 for less structured data. The rise of AI agents—which query warehouses at scales far beyond human users—has further amplified the need for low-latency, cost-efficient processing. In March 2026, Redshift already improved BI dashboard and ETL performance by accelerating new queries up to 7 times. Now, with RG instances, Redshift takes another major step forward.

Meet the New RG Instances

The RG instance family is built on AWS Graviton2 processors, delivering a compelling price-performance ratio. These instances run data warehouse workloads up to 2.2x faster than comparable RA3 instances, while reducing per-vCPU costs by 30%. But speed isn't the only advantage. RG instances come with an integrated data lake query engine that allows you to run SQL analytics across both warehouse tables and your Amazon S3 data lake from a single engine. For Apache Iceberg tables, performance is up to 2.4x faster than RA3; for Apache Parquet, it's up to 1.5x faster. This blend of speed and unified querying makes RG instances ideal for modern analytics and AI workloads.

Performance Gains and Cost Efficiency

When comparing RG instances to the current RA3 line, the improvements are clear. For example, the ra3.xlplus instance is replaced by rg.xlarge, offering the same vCPU and memory for small departmental analytics. Larger workloads benefit from a vCPU increase from 12 to 16 and memory from 96 GB to 128 GB in the rg.4xlarge—a 1.33:1 ratio. This translates to better throughput for standard production workloads. To estimate your potential savings, use the AWS Pricing Calculator with your specific workload patterns.

Integrated Data Lake Query Engine

One of the standout features of RG instances is the built-in data lake query engine, enabled by default. You no longer need separate tools or external services to query data on Amazon S3. This unified approach reduces operational complexity and total analytics costs for businesses running combined warehouse and lake workloads. Whether your data is in Iceberg or Parquet format, you get optimized performance without moving data.

Ideal for AI and Analytics Workloads

The combination of high throughput and low latency makes RG instances exceptionally well-suited for today's agentic AI workloads, where autonomous agents execute numerous concurrent queries. BI dashboards, near-real-time analytics, and ETL pipelines also benefit from improved response times. The integrated engine ensures that AI agents can access both warehouse and lake data seamlessly.

Migrating to RG Instances

Getting started with Amazon Redshift RG instances is straightforward. You can launch new clusters or migrate existing ones using the AWS Management Console, AWS CLI, or API. The integrated data lake query engine is turned on automatically, so no extra configuration is required. For detailed migration steps, refer to the official documentation.

Conclusion

Amazon Redshift RG instances represent a significant step forward in cloud data warehousing. By combining Graviton-based performance—up to 2.2x faster than RA3—with a unified data lake query engine, they address the needs of both traditional analytics and AI-driven workloads. Lower per-vCPU costs and simplified operations make them a compelling choice for organizations looking to optimize their data infrastructure. Explore the migration guide and start evaluating how RG instances can transform your analytics pipeline today.