The Evolution of Amazon Redshift
Since its launch in 2013, Amazon Redshift has been a game-changer for cloud data warehousing, delivering enterprise-grade performance at a fraction of on-premises costs. Over the years, the service has evolved through multiple architectural leaps—from dense compute nodes to RA3 instances, and from provisioned clusters to Amazon Redshift Serverless. Each generation has brought faster, cheaper, and more efficient query execution. Today, organizations manage exploding data volumes and increasingly complex analytics, often combining structured warehouse tables with cost-effective data lake storage. The rise of AI agents, which generate massive query loads, adds further pressure on operational costs and performance. In response, Redshift has consistently doubled down on its core strengths. For instance, in March 2026, Redshift accelerated BI dashboards and ETL workloads by boosting new query speeds up to 7 times, dramatically improving response times for low-latency SQL operations.
Introducing RG Instances: Built on AWS Graviton
Today, Amazon Redshift announces RG instances, a new instance family powered by AWS Graviton processors. These instances are engineered to deliver a step-change in price and performance for data warehousing and analytics workloads. Key highlights include:
- Up to 2.2x faster data warehouse performance compared to RA3 instances.
- 30% lower price per vCPU, significantly reducing total cost of ownership.
- Integrated data lake query engine that unifies analytics across warehouse tables and Amazon S3 data lakes using a single SQL engine.
- Performance for Apache Iceberg tables is up to 2.4x faster than RA3, and for Apache Parquet up to 1.5x faster.
This combination makes RG instances ideal for the high query volumes and low-latency demands of modern analytics and agentic AI workloads.
Performance and Cost Improvements
The Graviton-based architecture allows Redshift to achieve remarkable efficiency gains. For mixed workloads—combining traditional data warehouse queries with data lake exploration—RG instances drastically reduce processing time and expense. The integrated query engine eliminates the need for separate systems, simplifying operations and cutting costs further.
Comparison with RA3 Instances
The table below maps current RA3 instances to their recommended RG counterparts:
| Current RA3 Instance | Recommended RG Instance | vCPU | Memory (GB) | Primary Use Case |
|---|---|---|---|---|
| ra3.xlplus | rg.xlarge | 4 | 32 | Small cluster departmental analytics |
| ra3.4xlarge | rg.4xlarge | 12 → 16 (1.33:1) | 96 GB → 128 GB (1.33:1) | Standard production workloads, medium data volumes |
These upgrades offer more vCPUs and memory at a lower price point, enabling faster queries and better concurrency. For accurate savings estimates, AWS recommends using the AWS Pricing Calculator with your specific workload patterns.
Use Cases: From BI to AI Agents
RG instances excel in scenarios where speed and cost efficiency are critical:
- Business Intelligence (BI) dashboards: Deliver sub-second query responses even with complex aggregations.
- ETL pipelines: Accelerate data transformation and loading processes.
- Near-real-time analytics: Support streaming data ingestion and ad-hoc analysis.
- Autonomous AI agents: Handle high query volumes from goal-seeking agents without spiraling costs.
The integrated data lake engine also enables seamless querying across warehouse and data lake, reducing the need for data movement and separate query engines.
Getting Started with RG Instances
You can launch new clusters or migrate existing ones through the AWS Management Console, AWS CLI, or AWS API. The integrated data lake query engine is enabled by default, so no extra configuration is needed. Simply choose an RG instance type, and Redshift automatically optimizes query execution across warehouse and S3 data lakes. For existing RA3 users, migration is straightforward—review the recommended instance mapping above and use the console to resize.
With RG instances, Amazon Redshift continues to deliver on its promise of making data analytics faster, cheaper, and simpler—whether the users are humans or AI agents.