Amazon Redshift RG Instances: Graviton-Powered Data Warehouse with Integrated Data Lake Queries

Welcome to our deep dive into the latest innovation from Amazon Redshift. In March 2026, Redshift already accelerated new queries by up to 7x, but now the bar has been raised again. The new RG instances, powered by AWS Graviton processors, combine a high-performance data warehouse with a native data lake query engine. This Q&A covers everything you need to know—from performance gains and cost savings to real-world use cases and migration steps.

What is Amazon Redshift and how has it evolved over time?

Amazon Redshift has been redefining cloud data warehousing since 2013, delivering enterprise-grade performance at a fraction of on-premises costs. Over the years, each architectural leap—from dense compute nodes to RA3 instances and from provisioned clusters to Redshift Serverless—has made queries cheaper, faster, and more efficient. As data volumes exploded and analytics demands grew more complex, organizations began using both structured data warehouse tables and cost-effective data lakes. The rise of AI agents further multiplied query loads, magnifying operational costs. Redshift responded by doubling down on performance. For example, in March 2026, it sped up new queries by up to 7 times for BI dashboards and ETL workloads. Now, with RG instances based on AWS Graviton, Redshift delivers yet another generational leap—improving speed by up to 2.2x over RA3 while cutting price per vCPU by 30%.

What are the new Amazon Redshift RG instances and what benefits do they offer?

The Amazon Redshift RG instance family is built on AWS Graviton processors, combining raw compute performance with an integrated data lake query engine. Compared to current RA3 nodes, RG instances run data warehouse workloads up to 2.2x faster and cost 30% less per vCPU. They are ideally suited for high-volume, low-latency analytics—whether driven by human queries or autonomous AI agents. The integrated engine allows you to run SQL analytics across both warehouse tables and Amazon S3 data lakes from a single system, eliminating the need for separate query engines. This reduces total analytics costs and simplifies operations. For example, RG instances deliver up to 2.4x faster performance for Apache Iceberg and up to 1.5x faster for Apache Parquet compared to RA3. These improvements are critical for modern workloads like near-real-time BI dashboards, ETL pipelines, and agentic AI applications that demand rapid responses at scale.

How do RG instances compare to existing RA3 instances?

The transition from RA3 to RG is straightforward, with direct instance mappings based on vCPU and memory. For small departmental analytics, the ra3.xlplus (4 vCPU, 32 GB) is replaced by the rg.xlarge with identical specs. For standard production workloads, the ra3.4xlarge (12 vCPU, 96 GB) maps to the rg.4xlarge (16 vCPU, 128 GB)—a 33% increase in both compute and memory. This upgrade delivers a 1.33:1 gain in resources while lowering per-vCPU costs by 30%. Overall, you can expect up to 2.2x faster query performance across diverse workloads. The integrated data lake query engine further amplifies value when working with data lake formats like Iceberg and Parquet. To estimate your specific savings, use the AWS Pricing Calculator with your workload patterns. You can migrate existing clusters or launch new ones via the AWS Management Console, CLI, or API.

How does the integrated data lake query engine work and what performance improvements does it provide?

The integrated data lake query engine within RG instances is enabled by default. It unifies querying across your Amazon Redshift warehouse tables and Amazon S3 data lake using a single SQL engine. This eliminates the overhead of moving data or maintaining separate analytics tools. Performance gains are substantial: for Apache Iceberg tables, RG instances are up to 2.4x faster than RA3; for Apache Parquet, they are up to 1.5x faster. These improvements stem from Graviton's efficient processing and optimizations in the query engine that reduce latency and increase throughput. The engine supports all standard SQL analytics, making it ideal for BI dashboards, ETL jobs, and AI agent queries that frequently hit both warehouse and data lake sources. By handling both environments in one engine, you gain faster insights at lower cost, while simplifying your data architecture.

How can organizations get started with Amazon Redshift RG instances?

Getting started is simple. You can launch new clusters directly from the AWS Management Console, or use the AWS Command Line Interface (CLI) or API. The integrated data lake query engine comes pre-enabled, so no additional configuration is needed. For existing RA3 clusters, migration is equally straightforward: you can resize or restore snapshots to RG instance types. Amazon recommends using the AWS Pricing Calculator to estimate cost savings based on your specific workload. The migration can be performed with minimal downtime, and you can test performance in a staging environment before switching production. Once on RG instances, you'll immediately benefit from up to 2.2x faster queries and 30% lower costs per vCPU. For detailed instructions, refer to the Amazon Redshift Management Guide.

Who would benefit most from using Amazon Redshift RG instances?

Organizations running high-volume, low-latency analytics workloads will see the greatest benefits. This includes business intelligence (BI) dashboards that require sub-second response times, ETL pipelines processing large datasets, and near-real-time analytics applications. The emergence of autonomous AI agents—which query data warehouses at scales far exceeding human usage—makes RG instances especially valuable. These agents need the performance to handle thousands of concurrent queries without spiking costs. The integrated data lake query engine also appeals to companies that store diverse datasets in S3 and want to analyze them together with warehouse data without duplicating storage or managing multiple engines. In short, if your analytics combine structured warehouse tables with data lake formats like Iceberg or Parquet, and you demand speed, cost efficiency, and operational simplicity, RG instances are the ideal choice.